Compton-thick AGN in the NuSTAR Era VI: The Observed Compton-thick Fraction in the Local Universe

The MYTorus model of Murphy & Yaqoob (2009) assumes an isotropic X-ray emission within a uniform, neutral (cold) torus. The half-opening angle of the torus is fixed to 60°.

The MYTorus model is decomposed into three different components: an absorbed line-of-sight emission, a reflected continuum, and a fluorescent line emission. These components are linked to each other with the same normalization, absorbing column density (the model yields the equatorial column density of the torus, N_H,eq) and inclination angle θ_i. The inclination angle is measured from the axis of the torus, so that θ_i = 0° represents a face-on AGN, and θ_i = 90° an edge-on one.

The line-of-sight column density of the torus can be obtained from the equatorial value as

$$\begin{eqnarray}&&{N}_{{\rm{H}},\mathrm{los}}={N}_{{\rm{H}},\mathrm{eq}}\times {\left(1\mbox{--}4\times \cos {\left({\theta }_{{\rm{i}}}\right)}^{2}\right)}^{1/2}\end{eqnarray} \tag{ 1 }$$

and the average torus column density is determined by the given configuration as N_H,av = π/4 × N_H,eq, and cannot be decoupled (i.e., fit separately) from the line-of-sight value.

Both the reflected continuum and line emission can be weighted via two additional constants, A_S and A_L, respectively. When left free to vary, these can account for differences in the actual geometry (compared to the specific model assumptions used in the original calculations) and time delays between direct, scattered, and fluorescent line photons.

In XSPEC, this model configuration is as follows:

$$\begin{eqnarray}&&\begin{array}{l}\mathrm{model}=C\ast \mathrm{phabs}\ast \\ (\mathrm{mekal}+\mathrm{mytorus}\_\mathrm{Ezero}\_{\rm{v}}00{\rm{.}}\mathrm{fits}\ast \mathrm{zpowerlw}\\ \ \ +\ {A}_{S}\ast \mathrm{mytorus}\_\mathrm{scatteredH}500\_{\rm{v}}00{\rm{.}}\mathrm{fits}\\ \ \ +\ {A}_{L}\ast \mathrm{mytl}\_{\rm{V}}000010\mathrm{nEp}000{\rm{H}}500\_{\rm{v}}00{\rm{.}}\mathrm{fits}\\ \ \ +\ {F}_{{\rm{s}}}\ast \mathrm{zpowerlw}),\end{array}\end{eqnarray} \tag{ 2 }$$

where C is a cross-calibration constant between different instruments, or a cross-normalization constant between different observations.

The MYTorus model can also be used in decoupled configuration (Yaqoob 2012), so as to better represent the emission from a clumpy torus. While the model has the exact same assumptions, a better description of the data is possible when decoupling the line-of-sight emission from the reflection component. That is, N_H,los and N_H,av are not fixed to the same value, allowing the flexibility of having a particularly dense line of sight in a (still uniform) Compton-thin torus, or vice versa.

In this configuration, the line-of-sight inclination angle is always fixed to θ_i = 90°, and two reflection and line components are included, one with θ_i = 90° (forward scattering) and weighted with A_S,L90 and one with θ_i = 0° (backward scattering) and weighted with A_S,L0. Note that in this configuration θ_i is no longer a variable, although the ratio between A_S,L0/A_S,L90 can give a qualitative idea of which direction reflection predominantly comes from, and therefore of the relative orientation of the AGN. ¹²

In XSPEC this model configuration is as follows:

$$\begin{eqnarray}&&\begin{array}{l}\mathrm{model}=C\ast \mathrm{phabs}\ast \\ (\mathrm{mekal}+\mathrm{mytorus}\_\mathrm{Ezero}\_{\rm{v}}00{\rm{.}}\mathrm{fits}* \mathrm{zpowerlw}\\ \ \ +\ {A}_{S,0}\ast \mathrm{mytorus}\_\mathrm{scatteredH}500\_{\rm{v}}00{\rm{.}}\mathrm{fits}\\ \ \ +\ {A}_{L,0}\ast \mathrm{mytl}\_{\rm{V}}000010\mathrm{nEp}000{\rm{H}}500\_{\rm{v}}00{\rm{.}}\mathrm{fit}s\\ \ \ +\ {A}_{S,90}\ast \mathrm{mytorus}\_\mathrm{scatteredH}500\_{\rm{v}}00{\rm{.}}\mathrm{fits}\\ \ \ +\ {A}_{L,90}\ast \mathrm{mytl}\_{\rm{V}}000010\mathrm{nEp}000{\rm{H}}500\_{\rm{v}}00{\rm{.}}\mathrm{fits}\\ \ \ +\ {F}_{s}\ast \mathrm{zpowerlw}){\rm{.}}\end{array}\end{eqnarray} \tag{ 3 }$$

We fix A_S,90 = A_L,90 and A_S,0 = A_L,0. In the default MYTorus decoupled scenario, we freeze all these constants to unity. However, we also include the results of thawing them both, which in some cases improves fit quality. We call this third model MYTorus decoupled free.

The other model used in this work is borus02 (Baloković et al. 2018). This model also assumes a uniform torus, but the opening angle is not fixed, and different geometries can be considered through the covering factor, C_F parameter (C_F ∈ [0.1, 1]). The model only includes a reflection component, which contains both the continuum and lines, meaning that an absorbed line-of-sight component must be added.

By default, we use this model in a decoupled configuration, with N_H,los and N_H,av set to vary independently. An advantage of this model, aside from including a variable covering factor, is that θ_i (with θ_i ∈ [18°–87°]) can still be fitted in a decoupled configuration. borus02 also includes a high-energy cutoff and iron abundance as free parameters, although we freeze them to 500 keV (consistent with the default setting in MYTorus) and 1, respectively, due to our inability to constrain them. We note that some works estimate lower (∼200 keV) high-energy cutoffs (e.g., Ricci et al. 2017; Ananna et al. 2020). However, the most recent work of Baloković et al. (2020) focuses on the local obscured AGN population (i.e., more similar to our sample properties) and places the average cutoff at ∼300 keV. They also show that systematic uncertainties allow for a relatively wide range, which marginally includes 500 keV. Note, also, that with NuSTAR data reaching up to 25–55 keV energies, the different high-energy cutoff values do not impact our results. CGCG 475-040 is the exception to this rule, as it required a very low high-energy cutoff to adequately fit the data.

In XSPEC this model configuration is as follows:

$$\begin{eqnarray}&&\begin{array}{l}\mathrm{model}=C\ast \mathrm{phabs}\ast (\mathrm{mekal}\\ \ \ +\ \mathrm{borus}02\_{\rm{v}}170323{\rm{a}}{\rm{.}}\mathrm{fits}+\mathrm{zphabs}\ast \mathrm{cabs}\ast \mathrm{zpowerlw}\\ \ \ +\ {F}_{s}\ast \mathrm{zpowerlaw}),\end{array}\end{eqnarray} \tag{ 4 }$$

where zphabs and cabs are the photoelectric absorption and Compton scattering, respectively, applied to the line-of-sight component.

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.03}_{-0.24}^{+0.40}$ based on XMM-Newton and Swift-BAT observations. There are no optical classifications on its activity type, but its optical spectrum (Jones et al. 2009) does not present any broad emission lines. Results of the fitting can be found in Table 2.

This source is well fitted by all models except for MYTorus decoupled. Note that, generally, MYTorus decoupled provides a better fit than the coupled version. This is likely because, for this particular source, there is no contribution coming from face-on reflection, and the mentioned model assumes A_S0 = A_S90 = 1. Indeed, MYTorus decoupled free is a better fit, showing a clear predominance of forward reflection, which agrees with the borus02 inclination angle being large (cos(θ_i ) <0.3).

All models are in agreement that this source is observed through a Compton-thin line-of-sight (N_H,los = 0.47−0.67 ×10²⁴ cm⁻²), while the average torus material is denser, and even Compton -thick (N_H,av = 0.71–2.42 × 10²⁴ cm⁻²). According to the borus02 best-fit model, this Compton-thick torus would be geometrically thin, with a relatively small covering factor (${C}_{{\rm{F}}}={0.21}_{-0.09}^{+0.32}$). The photon index lies in the range Γ =1.40–1.83, when considering all the different models.

MCG 07-03-007 ¹³ was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.18}_{-0.35}^{+0.12}$ based on Swift-XRT and Swift-BAT observations. The source is optically classified as an Sy2 (Baumgartner et al. 2013). Results of the fitting can be found in Table 5.

All models are in good agreement for the description of the source. It is marginally Compton thin in the line-of-sight (N_H,los = 0.73–0.97 × 10²⁴ cm⁻²), with Γ ∼ 1.8 and a Compton-thick torus. The results of MYTorus decoupled free have larger uncertainties, likely due to the fact that the addition of two more parameters is not required by the fit. borus02 gives a best fit with a covering factor of 0.6, just barely intercepted by the line of sight.

This source was selected as a candidate CT-AGN based on our own Swift-XRT and Swift-BAT analyses, which is unreported in any previous publications. Optically, it is classified as an Sy2 (Baumgartner et al. 2013). Results of the fitting can be found in Table 6.

This source is clearly best fit by MYTorus decoupled free and borus02, with remarkably similar results, which (except for the photon index) do not differ significantly from the MYTorus decoupled best-fit model. According to our analysis, this source is borderline Compton thick in the line of sight, and Compton thick in the average torus material (N_H,los = 0.92–1.09 × 10²⁴ cm⁻², N_H,av = 2.86–4.64 × 10²⁴ cm⁻²). borus02 can constrain the covering factor and inclination angle with high accuracy thanks to the clear dominance of the reflection component. For this source, the dominance of forward reflection (according to MYTorus decoupled free) would lead us to believe that the source has a large inclination angle, but borus02 results place it at θ_i ∼ 30°. It could be that, given the large covering factor (${C}_{{\rm{F}}}={0.97}_{-0.03}^{+0.02}$), most of the reflection comes through the torus, which MYTorus interprets as 90° reflection, regardless of the actual direction.

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.03}_{-0.10}^{+0.30}$ based on Swift-XRT and Swift-BAT observations. Optically, it is classified as an Sy2 (Baumgartner et al. 2013). Results of the fitting can be found in Table 7.

For this source, MYTorus decoupled free and borus02 are in strong agreement, fitting it as Compton thick in the line-of-sight (N_H,los ∼ 1.45 × 10²⁴ cm⁻²), with a lower average torus column density (N_H,av ∼ 0.33 × 10²⁴ cm⁻²). With an estimated covering factor of only C_F = 0.15, this source is likely to have a very patchy torus. In this scenario, the low covering factor should not be interpreted geometrically (i.e., like in a thin disk-like torus) but rather physically, meaning that the surrounding clouds obscure only a small fraction of the available volume.

Contrary to this, MYTorus decoupled gives a different estimation for all the significant parameters, with a harder photon index, Compton-thin line of sight and Compton-thick torus. However, the fit is statistically worse, and the MYTorus decoupled free results point toward a strong dominance of only forward reflection (agreeing with the borus02 edge-on viewing angle). This leads us to favor the former scenario, given how MYTorus decoupled is limited by forcing A_S,L90 = A_S,L0 = 1.

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.29}_{-0.30}^{-}$ ¹⁴ based on XMM-Newton and Swift-BAT observations. Optically, it is classified as an Sy2 (Véron-Cetty & Véron 2006). Results of the fitting can be found in Tables 8 and 9, the latter of which includes the second XMM-Newton observation, taken from the archive.

In the case of MRK 622, the MYTorus model does not show a statistical improvement when adding additional free parameters. All models place this source as having a Compton-thick torus (N_H,av = 0.63–3.11 × 10²⁴ cm⁻²), while having a much lower line-of-sight column density (N_H,los = 0.15–0.29 ×10²⁴ cm⁻²). There is a disagreement between the best-fit photon index value between the MYTorus model and the borus02 model, with the first set at ${\rm{\Gamma }}={1.54}_{-}^{+0.14}$, and the second at ${\rm{\Gamma }}={1.74}_{-0.19}^{+0.17}$. The determination of torus properties, such as covering factor and opening angle, is made difficult by the fact that the reflection component is subdominant with respect to the line of sight (hence, the unconstrained values). This is also likely to be the reason for MYTorus decoupled free to not show a statistical improvement of fit, as the added parameters model reflected emission.

When adding the archived data, we introduce a cross-normalization constant, C, and a different line-of-sight hydrogen column density, N_H,los2, and leave both free to vary. The addition does not result in significant changes or incremented agreement between the different models. However, leaving N_H,los2 free to vary results in a best-fit value of N_H,los2 = 0.39–0.69 × 10²⁴ cm⁻², which is incompatible within the errors with the best-fit value for N_H,los, for all models (see Figure 2). Therefore, we conclude that this source presents line-of-sight N_H variability at different epochs.

Zoom In Zoom Out Reset image size

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.05}_{-0.22}^{+0.35}$ based on XMM-Newton and Swift-BAT observations. Optically, it is classified as an Sy2 (Lin et al. 2012). Results of the fitting can be found in Tables 10 and 11, the latter of which includes the second XMM-Newton observation, taken from the archive.

All models are in agreement that this source has a Compton-thick line-of-sight (N_H,los = 1.42–3.16 × 10²⁴ cm⁻²) within a Compton-thin torus (N_H,av = 0.30–0.55 × 10²⁴ cm⁻²). The range of photon index values is relatively large (Γ = 1.53–2.11), although both MYTorus decoupled and decoupled free are compatible with the borus02 results within errors. The ratio between A_S,90 and A_S,0 suggests a predominance of forward reflection, which is compatible with the observation angle derived by borus02 , θ_i ∼ 75°.

When adding the archived observation, the data quality did not allow constraining N_H,los2, and its value was compatible with that of N_H,los. Therefore, the results presented in Table 11 have them fixed to be the same value. Both this and the fact that C is compatible with 1 makes us conclude that this source does not present variability between the two analyzed observations.

The addition of this second set of XMM-Newton data improves the overall agreement between the three models with good fitting statistics, and in particular between MYTorus decoupled free and borus02 . The qualitative description of the results would remain the same, with reduced uncertainty; N_H,los = 1.42−2.56 ×10²⁴ cm⁻², N_H,av = 0.34−0.63 × 10²⁴ cm⁻², and Γ = 1.57−1.84.

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.79}_{-0.40}^{-}$ based on Swift-XRT and Swift-BAT observations. It is optically classified as an Sy2 in Bird et al. (2007). This source has an additional XMM-Newton+NuSTAR joint observation (PI: Marchesi, Obs IDs: 0852180801 and 60561045002) that we do not use due to an error in the data taking. Results of the fitting can be found in Table 12.

All models fit this source with a slightly soft photon index (Γ ∼ 2.0) and a Compton-thick line-of-sight, N_H,los = 1.12–3.21 ×10²⁴ cm⁻² (ignoring the MyTorus unconstrained result, as the statistics do not justify the addition of two extra free parameters). Also in agreement, they place the average torus column density to be slightly lower, N_H,av = 0.49–1.79 × 10²⁴ cm⁻². Again, borus02 gives a best fit with a covering factor of 0.6, just barely intercepted by the line of sight.

This source was first reported as a CT-AGN candidate by Ricci et al. (2015), who obtained a value of log ${N}_{{\rm{H}}}={24.20}_{-0.20}^{+0.30}$ based on Swift-XRT and Swift-BAT observations. It is optically classified as an Sy2 in Parisi et al. (2014). Results of the fitting can be found in Table 13.

This source is best fit with a rather soft photon index, Γ = 1.9–2.60, and a large value of the average hydrogen column density when using the MYTorus model. The borus02 best fit of the NuSTAR+XMM-Newton data has two possible configurations fitting the data equally well: (1) a soft photon index, with a dense torus that has a large (∼90%) covering factor, and is viewed at a small inclination angle; (2) a photon index frozen to 1.8, with a very patchy torus (low average column density and covering factor, yet Compton thick in the line of sight). As both options have the same reduced χ² we cannot say which model is superior.

To try and disentangle this degeneracy, we added Swift-BAT data to the spectrum, ¹⁵ as shown in Figure 13. We used borus02 to fit the data, as it is the only model providing an estimate for the torus covering factor, which we are interested in constraining within the two possible options mentioned above. With the addition of the BAT data, the best-fit model favors a scenario with a very dense torus of large covering factor, through which we observe the AGN through an underdense region (since ${N}_{{\rm{H}},\mathrm{los}}={1.60}_{-0.15}^{+0.23}\times {10}^{24}$ cm⁻²). Although the average torus density is capped at the maximum possible value (log N_H,av = 25.5), we note that it is compatible with being only a factor ∼1.5 larger than that of the line of sight.

Interestingly, the BAT data can be adequately fit only when considering a cross-normalization factor between the NuSTAR+XMM-Newton data and the BAT data ($C={2.38}_{-0.48}^{+0.49}$). This implies our joint observation took place in a low-flux state of the source. It is also necessary to leave the high-energy cutoff free to vary, for which we obtain ${E}_{\mathrm{cut}}={21.0}_{-}^{+17.7}$ keV. This value, while being low, is not unprecedented (see the cutoff energy distribution of Swift-BAT sources; Ricci et al. 2017; Ananna et al. 2019). However, we caution that such a low high-energy cutoff can be spurious when N_H is high and data quality is not exceptional (see the discussion in, e.g., Baloković et al.2020).

All models classify the source as Compton thick, with N_H,los > 10²⁴ cm⁻². MYTorus decoupled free, despite having a Compton-thick best-fit value, is also compatible with a Compton-thin scenario within errors. However, we note that this model is statistically worse than MYTorus decoupled despite having two additional free parameters. This likely means MYTorus decoupled free has too many free parameters, which increases degeneracy and results in less reliable results.

Ricci et al. (2015) provided a list of CT-AGN candidates in the 70 month BAT catalog, based on joint BAT and soft X-rays analysis (the best available data out of XMM-Newton, Chandra, Suzaku, Swift-XRT, and Advanced Satellitle for Cosmology and Astrophysics; ASCA). Out of a total of 55, 50 fall within z < 0.05. Based on the Palermo 100 catalog (Cusumano et al. 2015; A. Segreto et al. 2021, in preparation), other works (Marchesi et al. 2017a, 2017b; R. Silver et al. 2021 in preparation) have added to the list of possible CT-AGN within the BAT catalog. This comes to a total of 63 CT-AGN candidates at z ≤ 0.05.

Marchesi et al. (2018) analyzed 38 of these sources using NuSTAR data, which is key to disentangling the degeneracy between the photon index, Γ, the line-of-sight column density, N_H,los, and the reflection component (which depends on the average column density, N_H,av, inclination angle, and torus covering factor), and confirmed the Compton-thick nature of 17 them (originally 20 in Marchesi et al. (2018), but the reanalysis performed by Zhao et al. (2021) detailed in Section 5.2 reclassified three of them into Compton thin). Further works have brought the number of BAT-selected, confirmed CT-AGN at z ≤ 0.05, to a total of 29 (Koss et al. 2016; Oda et al. 2017; Georgantopoulos & Akylas 2019; Tanimoto et al. 2019; Kammoun et al. 2020; Zhao et al. 2020, 2021; Traina et al. 2021).

In this work, we report five additional NuSTAR-confirmed Compton-thick sources, bringing the current number to 32. This work thus represents an ∼19% increase of confirmed CT-AGN over the previous sample. The full list of NuSTAR-confirmed CT-AGN in the BAT catalog at z ≤ 0.05, along with the references to their analysis, can be found at our website. ¹⁶

According to the latest version of the Palermo BAT catalog (100 month catalog, A. Segreto et al. 2021, in preparation), there is a total of 414 BAT-detected AGN within z ≤ 0.05. We note that we are including galaxies lacking an optical classification as AGN (i.e., galaxy, galaxy in pair, galaxy in group, emission-line galaxy, and infrared galaxy), given how their bright emission at >15 keV is difficult to explain through other means. This implies that the number of confirmed CT-AGN within the volume-limited sample is still ∼8%, far from model predictions. The most recent estimate, that of Ananna et al. (2019), predicts, based on population synthesis models, a fraction of CT-AGN of ∼50%. Note, however, that this prediction is dependent on the flux of the sample considered. Our BAT sample is flux limited, and after applying the pertinent correction, should have a CT-AGN fraction between 27% and 38%, according to the model of Ananna et al. (2019)

We also note that our z < 0.05 is not necessarily complete. Indeed, we observe a significant trend with redshift of the CT-AGN fraction, which is higher (i.e., closer to predicted values) at lower redshifts. Figure 3 and Table 4 show the evolution of the observed CT-AGN fraction as a function of redshift, proving that indeed we can recover an ∼20% of CT-AGN in the lowest redshift bin. This value lies just below (when taking errors into account) the lower limit of the Ananna et al. (2019) estimations.

Zoom In Zoom Out Reset image size

These results point to the sample being incomplete even at the lowest possible redshift, likely a result of the BAT flux limit. CT-AGN at higher redshift are likely too faint to be detected by BAT in the first place. We note that the errors listed in Table 4 and shown in Figure 3 are purely statistical, and do not account for any bias or incompleteness/obscuration corrections. Such predictions are nontrivial and are left for a future study.

In Figure 4, we show the sources with NuSTAR data targeted by our group as part of the CCTAGN project, our effort to characterize CT-AGN in the Local Universe. Out of the 48 objects analyzed by our group, 24 are found to be CT-AGN, which represents a success rate of ∼50%. This result showcases the need for NuSTAR data to confirm CT-AGN candidates, as all sources shown in the figure were compatible with being Compton thick based on soft X-rays and BAT data. It is possible that, without NuSTAR data and using simpler, phenomenological models, the selection based on BAT + soft X-ray data cannot properly distinguish between N_H,los and N_H,av, resulting in misclassifications (all Compton-thin sources in this work have a Compton-thick torus). We note that using the phenomenological models on the high-quality data we present here would result in the same effect. However, one can only apply the self-consistent, complex models (that allow disentanglement between N_H,los and N_H,av) in a meaningful way if NuSTAR data is available. That is because the reflection component dominates at ∼20 keV, a range in which no other satellite is sensitive enough.

Zoom In Zoom Out Reset image size

The mentioned, phenomenological models, which do not take into account the decoupling between N_H,los and N_H,av, are typically used in population synthesis models. The spectral shape of a source that is Compton thin in the line of sight but has a large N_H,av might have a spectrum not so dissimilar from a source with a homogeneous Compton-thick torus, in which reflection is not taken into account in a self-consistent way. These sources, while not CT-AGN in the line of sight, might still contribute to the CXB in a relevant way. Therefore, we caution that our observed 20% ± 5% fraction of CT-AGN at z < 0.01 should be compared with the predictions of population synthesis models in a careful way.

Based on our results, all eight sources except for ESO 323-G023 are incompatible, at 90% significance, with having the same line of sight and average torus column densities. Even for ESO 323-G023, the two values are compatible only at the very limits of their error range.

In Figure 4, we plot the sources analyzed by our group, comparing their line of sight and average torus column densities. Originally, Marchesi et al. (2018) analyzed the 38 sources in their sample using borus02 with the inclination angle frozen to θ_i = 90° to obtain an estimate of N_H,av without leaving the parameter free to vary. Zhao et al. (2021) reanalyzed most of the sample (i.e., those sources with good-enough data quality) using the same methodology as described in this work. In Figure 4, we plot the sources as reanalyzed by Zhao et al. (2021) when available. ¹⁷ For those with insufficient data quality, we note that the determination of N_H,av should be taken as a rough estimate. On the other hand, a comparison of N_H,los for both analyses has shown little difference for most sources ¹⁸ Five additional sources are not included in the plot, due to having N_H < 10²³ cm⁻² (2MASXJ10523297+1036205, RBS 1037, MCG-01-30-041, B2 1204+34, and ESO 244-30, Marchesi et al. 2019). These sources, all originally CT-AGN candidates, highlight even further the importance of using NuSTAR to estimate N_H,los.

Figure 4 shows no strong correlation between one quantity and the other, and in fact, tend to fall far from the one-to-one relation. This is confirmed by our statistical analysis, which yields a Pearson correlation coefficient of ρ ≈ −0.02. ¹⁹ This means that sources that are Compton thick in the line of sight are no more likely to have a thicker torus than other sources. This result is in agreement with that found by Zhao et al. (2021), who analyzed a sample of ∼100 Compton-thin AGN in the Local Universe, which have high-quality NuSTAR and soft X-ray data, finding that N_H,av is similar at different N_H,los.

Furthermore, Figure 4 shows how most of the analyzed sources have significant differences between their estimated N_H,los and N_H,av values, which is a strong argument in favor of a patchy torus scenario. Observations both in the X-ray and in the near-IR band have already suggested that the structure of the obscuring material surrounding the supermassive black hole is, not surprisingly, more complex than a simple, homogeneous torus. Soft X-ray monitoring of AGN has shown variability in N_H,los (e.g., Risaliti et al. 2002; Elvis et al. 2004; Markowitz et al. 2014). Infrared observations of AGN torus emission also support this scenario (Ramos Almeida et al. 2014). Despite this vision being largely accepted, there is still a small sample of sources for which N_H variability has been confirmed, and other studies challenge this scenario after finding no variability in large samples (e.g., Laha et al. 2020, found no significant N_H variability in 13/21 analyzed Sy2's).

In this work we also find N_H,los variability for MRK 622 (in observations taken 16 yr apart), and we found none for NGC 6552 (in observations taken 17 yr apart). However, a complete analysis of column density variability and torus cloud distribution is beyond the scope of this work, and will be reported elsewhere. In order to draw stronger conclusions, we plan on targeting promising candidates in our sample (i.e., those with large N_H,los/N_H,av differences, or those with multiple observations), and analyze them with models based on clumpy torus distributions (e.g., UXClumpy; Buchner et al. 2019). A similar idea was proposed by Yaqoob et al. (2015). We leave this analysis for future work.

Generally, MYTorus and borus02 are in good agreement in their parameter estimation, particularly on their qualitative description of the source. That is, the models agree on Compton-thin versus Compton-thick classifications, both for N_H,los and N_H,av. They are also generally compatible in their photon index estimation (except for MRK 622 and ESO 112-G006), agreeing within errors for most of the sources.

This is not true when using MYTorus in a coupled configuration, as it tends to present strong disagreements with the photon index estimation, as well as systematically worse fits. Note, however, that the line-of-sight column density estimation is generally in agreement with that of the other models.

MYTorus decoupled and borus02 also agree in their qualitative description of the relative inclination of the source, with the ratio between A_S,90 and A_S,0 (showing predominance for forward or backward scattering) being consistent with the inclination angle and covering factor as estimated by borus02.

A detailed comparison of the MYTorus and borus02 performances can be found in Marchesi et al. (2019).