Revisiting the spectral energy distribution of I Zw 1 under the CaFe Project

The first-ionized state of iron (Fe ii) emission is observed from the ultraviolet to the near infrared (NIR) and acts as one of the main coolants of the broad-line region (BLR, Marinello et al., 2016; Marziani et al., 2018) and manifests as a pseudo-continuum owing to the many, blended multiplets over a wide wavelength range (see Verner et al., 1999; Kovačević et al., 2010, and references therein). It is a key parameter in (1) the classification of Type-1 AGNs in the context of the main sequence of quasars (Boroson and Green, 1992; Marziani et al., 2018; Panda et al., 2019), and (2) to realize an updated radius-luminosity relation wherein the inclusion of the strength of the Fe ii¹¹1this is well known as the R_FeII parameter which is the ratio of the integrated Fe ii emission within 4434-4684Å  to the broad H$\beta$ emission. relates to the accretion rate of the source. Seminal works led by Boroson and Green (1992); Verner et al. (1999); Sigut and Pradhan (2003) and others encapsulate the ‘yet to be complete’ understanding of the physics of the Fe ii line formation. The Fe ii pseudo-continuum can be modelled appropriately with an 8-dimensional parameter space, encompassing the full diversity of Type-1 AGNs as has been concluded from prior works²²2We refer the readers to the PhD Thesis for a comprehensive overview on this issue. A PDF version of the thesis can be accessed using the following link.. These 8 parameters consist of the fundamental black hole (BH) and BLR properties, namely (1) the Eddington ratio ($L\mathrm{{}_{bol}}$/$L\mathrm{{}_{Edd}}$), (2) the BH mass, (3) the shape of the ionizing continuum³³3The shape of the ionizing continuum is a generic term that is used to specify the distribution of the specific photon energy (in units of $\nu$F_ν or $\lambda$F_λ, or in corresponding luminosity units) as a function of frequency ($\nu$) or wavelength ($\lambda$). The term signifies the underlying continuum originating from the very central part of the BH, i.e., the thermally-radiating accretion disk and the Comptonized radiation from the hot/warm corona, that is incident on the BLR cloud. or the spectral energy distribution (SED), (4) the BLR local density, (5) the metal content in the BLR, (6) the velocity distribution of the BLR including turbulent motion within the BLR cloud⁴⁴4Mainly with the information of the BH mass and the velocity distribution of the BLR primarily influenced by the central gravitational potential of the BH, and under the assumption of the virial relation, we can derive the distance of the BLR cloud from the BH (i.e. $R\mathrm{{}_{BLR}}$). Thus, the two quantities - the velocity distribution and the $R\mathrm{{}_{BLR}}$ are closely connected., (7) the orientation of the source (as well as the BLR) with respect to the distant observer, and (8) the sizes of the BLR clouds.

However, the complex electronic structure of Fe ii owing to varied excitation mechanisms makes it difficult to model the atom ‘perfectly’. This opens up the possibility to search for viable alternatives. Past studies have suggested the existence of a zone shielded from the high-energy photons emanated by the central source and likely located in the outermost portion of the BLR (Joly, 1987; Dultzin-Hacyan et al., 1999; Rodr´ıguez-Ardila et al., 2002; Rodriguez-Ardila et al., 2012; Garcia-Rissmann et al., 2012; Marinello et al., 2016) with the presence of emission lines with very low-ionization potentials (IP$\sim$10 eV) such as the Ca ii triplet at $\lambda 8498,\lambda 8542,\lambda 8662$ (hereafter CaT) and O i $\lambda 8446$, in addition to the multiple permitted Fe ii transitions. The similarity in the location of the line production of these species suggests a common origin of these LILs, especially between the strengths of the Fe ii and the CaT⁵⁵5these strengths are estimated by normalizing the LILs emission to the broad H$\beta$ emission and are referred to as R_FeII and R_CaT, respectively. Like R_FeII which is the ratio of the optical Fe ii emission within 4434-4684Ånormalized to the H$\beta$ emission, the R_CaT is the ratio of the CaT emission normalized also to the same H$\beta$ emission.. This has been confirmed both from observational and photoionization studies in our recent works (Martínez-Aldama et al., 2015; Panda et al., 2020; Mart´ınez-Aldama et al., 2021b).

The combined importance of the two species has also been recognized in addition to our findings that the CaT being an effective proxy serves to be a better alternative to Fe ii-based $R\mathrm{{}_{H\beta}}-L_{5100}$ relation (Mart´ınez-Aldama et al., 2021a). This is crucial to address the scatter seen due to the inclusion of newer measurements and sources in the $R\mathrm{{}_{H\beta}}-L_{5100}$ relation specifically showing a deviation from the classical two-parameter $R\mathrm{{}_{H\beta}}-L_{5100}$ relation (Bentz et al., 2013). A large subset of these sources are noted to belong to the class of Narrow-line Seyfert Type-1 galaxies (NLS1s) that show shorter time delays, and hence a smaller radial distance of the onset of the BLR from the central continuum source ($R\mathrm{{}_{BLR}}$). These smaller $R\mathrm{{}_{BLR}}$ values show a marked deviation from the expected $R\mathrm{{}_{H\beta}}-L_{5100}$ relation, and addressing this problem is key to our understanding of how these systems evolve and if viable corrections to the classical relation can be made to utilize AGNs as “standardizable” cosmological candles. In addition to this issue, we find in Mart´ınez-Aldama et al. (2021b) that the ratio of the CaT to Fe ii (justifying the project name - CaFe) is an effective tracer of the chemical evolution of AGNs and can help us probe the co-evolution of the AGN and its host galaxy in more detail.

The dearth of observations in the NIR limits our current sample to $\sim$60 sources, but the increased availability of optical and NIR spectroscopic measurements, especially with the advent of the upcoming ground-based 10-metre-class (e.g. Maunakea Spectroscopic Explorer, Marshall et al., 2019) and 40 metre-class (e.g. The European Extremely Large Telescope, Evans et al., 2015) telescopes; and space-based missions such as the James Webb Space Telescope and the Nancy Grace Roman Space Telescope would further help to accentuate the strong correlation shown by these two ionic species.

In this short paper, we focus on a key issue of how the shape of the ionizing SED affects the production of these LILs - especially the Fe ii and CaT, and whether it leads to a substantial change in our existing results suggesting a common origin of these species. The paper is organized as follows - In Section 2 we outline the photoionization setup and the preparation of the new SED for the prototypical NLS1 - I Zw 1. In Section 3 we present the results from our analysis and discuss their implication on the existing connection between the two species. We summarize our findings in Section 4.

We apply the photoionization setup prescription that was demonstrated in Panda (2021, hereafter P21). We describe briefly the setup here - we perform a suite of CLOUDY (version 17.02, Ferland et al., 2017) models⁶⁶6N(${U}$) $\times$ N(${n_{H}}$) $\times$ N(Z) = 12$\times$11$\times$5 = 660 models by varying the mean cloud density, $10^{10.5}\leq n_{H}\leq 10^{13}\;(\rm{cm^{-3}}$, the ionization parameter, $-4.25\leq\log U\leq-1.5$, the metallicity, 0.1Z_⊙ $\leq$ Z $\leq$ 10Z_⊙, at a base cloud column density, $N_{\rm{H}}$ = $10^{24}$ cm^-2. The choice for the range for these physical parameters has been studied in detail in prior works (Panda et al., 2020; Panda, 2021) especially connected to the low-ionization emission lines (LILs), e.g. H$\beta$, Fe ii and CaT.

Our main focus in this paper is to highlight the role that the shape of the ionizing continuum (or the spectral energy distribution, i.e. SED) plays in modifying/constraining the physical parameter space for the effective emission of these LILs from the BLR. We assume that the region producing these LILs is dust-free, and the limit on the cloud column density restricts the region to effectively being free of additional scattering effects. In P21 (also in Panda et al., 2020, hereafter P20), we presented the results from our photoionization modelling that incorporated a SED for the prototypical NLS1 source I Zw 1 that was prepared using only photometric data points across a wide wavelength range ($\sim$0.3 mm to $\sim$1200 Å). We also incorporated in P20 an alternate SED which included photometric data points in the hard X-ray region ($\sim$3 Å), but this did not change our conclusions in P20 significantly, i.e. the parameter space in terms of the ionization parameter (U) and cloud local density (n_H) required to maximize the R_FeII and R_CaT remain unchanged.

Here, we extract the continuum from the high-resolution composite spectra for I Zw 1 using archival Hubble Space Telescope (HST) data (Bechtold et al., 2002) in the UV that is complemented with data in the optical (obtained using the 2.15m Complejo Astronomico El Leoncito - CASLEO) and in the NIR (obtained using the 3.2m NASA Infrared Telescope Facility - IRTF)⁷⁷7The optical and NIR spectrum were obtained and analyzed in Rodr´ıguez-Ardila et al. (2002); Riffel et al. (2006).. For the continuum points extraction, we automatically identify the emission lines, and select regions in the spectrum free of them to extract these points. A full description of the procedure can be found in a different work (Dias dos Santos et al. in prep.). The extracted continuum points are then supplemented with the photometric data points in the X-ray region and wavelengths above 2.5 $\mu$m from the previously used SED in P20 and P21. Figure 1 shows the comparison between the old SED from P20 and the new SED that is prepared in this work. We can appreciate the “Big Blue Bump” feature (Czerny and Elvis, 1987; Panda et al., 2018) in our new SED that is more prominent than the older one. This eventually leads to an excess of ionizing photons at the hydrogen ionization limit. Although we are aware that a direct interpolation between the last data point in the UV and the first data point in the X-ray regime is artificial and can lead to an unwarranted excess in the number of ionizing photons, as noticed in our previous work (P20), the hard X-ray photons do not directly interfere in the ionization of the LILs. These hard X-ray photons when coming into contact with the dusty torus (i.e. at distances $\sim$ 1 parsec from the central ionizing source), which has significantly higher optical depths, are known to eventually scatter producing low-energy photons but primarily in the mid to near IR due to the presence of dust (Horst et al., 2008; Padovani et al., 2017; Hönig, 2019). Thus, these hard X-ray photons have very little role to play in contributing in the ionization of these LILs.

Figure 2 shows the $\log{U}-\log{n_{H}}$ parameter space for the R_FeII (upper panels) and R_CaT (lower panels). The diagnostic plots on the left row are obtained with the new SED while those on the right are for the older SED (used in P20 and P21). We incorporate the prescription from Nenkova et al. (2008) to separate the dusty and non-dusty regime in the BLR, which has a form:

where, $\rm{R_{sub}}$ is the sublimation radius (in parsecs) computed from the source luminosity that is consistent for a characteristic dust temperature. This is a simplified version of the actual relation which, in addition to the source luminosity term, contains the dependence on the dust sublimation temperature and the dust grain size. We assume a dust temperature $T_{sub}$ = 1500 K, which has been found consistent with the adopted mixture of the silicate and graphite dust grains, and a typical dust grain size, a=0.05 microns. The dependence of the $\rm{R_{sub}}$ on the temperature is quite small - the exponent on the temperature term is -2.8. On the other hand, the dust grain size is a more complex problem, yet the value adopted is fair in reproducing the characteristic dust sublimation radius in our case (see Nenkova et al., 2008; Hönig, 2019, for more details). The sublimation radius, hence, is estimated using only the integrated optical-UV luminosity for I Zw 1. This optical-UV luminosity is the manifestation for an accretion disk emission and can be used as an approximate for the source’s bolometric luminosity. The bolometric luminosity⁸⁸8This bolometric luminosity value is quite similar to the value obtained by integrating the area under the curve in our new SED. of I Zw 1 is L${}_{\rm{bol}}\sim 4.32\times 10^{45}$ erg s^-1. This is obtained by applying the bolometric correction prescription from Netzer (2019) to I Zw 1’s optical monochromatic luminosity, $\rm{L_{5100}}\sim 3.48\times 10^{44}\;\rm{erg\;s^{-1}}$ (Persson, 1988). This uniquely sets the dust sublimation radius at $\sim 0.83\rm{pc}\;$(= $2.56\times 10^{18}\;$cm). Projecting this sublimation radius on the $\log{U}-\log{n_{H}}$ plane allows us to recover the non-dusty region that well represents the physical parameter space consistent with the emission from the BLR. This dust-filtering is applied to the models in a post-photoionization stage.

Focusing first on the upper panels in Figure 2, the R_FeII plots show a slight change in the location of the maximum, the new diagnostics suggest an ionization parameter that is about 0.25 dex higher and a shift by a similar factor is noticed in the local density, albeit reduced. Also, the value of the R_FeII obtained at the maximum is reduced by 20%. To assess the radial distances of the Fe ii emitting region, we use the formulation from P21, i.e.:

where, $R\mathrm{{}_{BLR}}$ is the distance of the emitting cloud (in cm) from the ionizing source which has a mean local density, ${n_{H}}$, and receives an ionizing flux that is quantified by the ionization parameter, U. $Q_{H}$ is the number of ionizing photons, which can be equivalently expressed in terms of the bolometric luminosity of the source per unit energy of a single photon, i.e. h$\nu$. Here, we consider the average photon energy, h$\nu$ = 1 Rydberg (Wandel et al., 1999; Marziani et al., 2015). The specific value of the bolometric luminosity corresponds to I Zw 1. Comparing the two plots in the upper panel (R_FeII based), the maximum R_FeII emitting location is shifted inwards by a factor 2 (for the P21-based SED plot, the maximum R_FeII is obtained for an ionization parameter, log U = -1.75, at a local cloud density, ${n_{H}}$ = 10^11.75 cm^-3. This returns a value for the $R\mathrm{{}_{BLR}}$ = 2.294$\times$10¹⁷ cm. On the other hand, for the new SED, the maximum R_FeII is obtained for a log U = -1.5, at ${n_{H}}$ = 10¹² cm^-3, which gives a $R\mathrm{{}_{BLR}}$ = 1.294$\times$10¹⁷ cm).

For the R_CaT, the location of the maximum value remains unchanged, although there is a slight increase in the net value in R_CaT with the new SED. The corresponding $R\mathrm{{}_{BLR}}$ location based on the maximum R_CaT location, is estimated to be about a factor 10 larger than the $R\mathrm{{}_{BLR}}$ for maximum R_FeII. We comment on this result in the next paragraphs.

As was inferred in P21, the location on the $\log{U}-\log{n_{H}}$ plane that leads to the maximum value for the flux ratios (R_FeII or R_CaT) do not agree in terms of their line equivalent widths when compared with observed estimates. As noticed through our simulations, for example, if the ratio of the EW(Fe ii) to EW(H$\beta$) is taken (i.e., the R_FeII), we can notice that at the location of the maximum value for R_FeII, the EW(Fe ii) is about 4Å, while for the EW(H$\beta$) at that same location the value obtained is about 2.5-3Å. This is in contradiction to the observed EWs measured from spectral fitting. In addition, such low EWs are almost at the limit of (or below) the observed spectral resolution. The regions wherein we find agreement both in terms of the flux ratios and the corresponding line EWs for these LILs are shifted towards lower ionization parameters (log U $\sim$ -3.0 and lower) for both these lines. We show the corresponding equivalent widths plots in Figure 3. The equivalent widths for Fe ii and H$\beta$ have been estimated using the continuum luminosity very close to the 5100Å (at 4885.36Å) and assuming a covering fraction of 20%, a value consistent with previous studies (Baldwin et al., 2004; Korista and Goad, 2001; Sarkar et al., 2021; Panda, 2021). For the CaT, we utilize a continuum closer to the line, i.e. 8329.68Å, and assume the same covering fraction. As can be noticed in the panels for these LILs in Figure 3, the EWs agreeable to observed estimates ($\sim$30-40Å) suggest a lowering in the log U, below -3.0, for the Fe ii emission. A similar shift is required for the CaT emission. This has already been noticed in P21 that at solar composition (without any microturbulence effects), the requested R_FeII and R_CaT values cannot be retrieved without agreeable EWs for these lines. An increase in the metal content (up to a factor 3-10) is required to match the observed flux ratios and line EWs for these LILs in I Zw 1. The similarities thus obtained in the $\log{U}-\log{n_{H}}$ parameter space brings the location of the emitting regions for the two species - Fe ii and CaT, almost to similar values of $R\mathrm{{}_{BLR}}$.

When compared with the observed flux ratios, we notice that the solar composition models shown here are insufficient to reproduce the R_FeII estimate by Persson (1988), i.e. 1.778$\pm$0.050 or the more recent estimate by Marinello et al. (2016), i.e. 2.286$\pm$0.199. As was concluded in P21 and also confirmed in Śniegowska et al. (2021), there is a need to increase the metal content to super-solar values, i.e. 3-10 Z_⊙, to push the Fe ii emission and thus the R_FeII estimate in perfect agreement with both these observed estimates. In contrary, the R_CaT estimates are successfully reproduced in these models with solar composition. R_CaT estimates for I Zw 1 are reported in Table 1 in P21: 0.513$\pm$0.130 (Persson, 1988) and 0.564$\pm$0.080 (Marinello et al., 2016).

To highlight the salient differences in incorporating the new SED in place of the existing SED from P21 for I Zw 1 we show $\Delta$R_FeII and $\Delta$R_CaT plots for our models in Figure 4. Locating the solutions in the $\log{U}-\log{n_{H}}$ plane where we get agreeable EWs for the two species, we notice that the new SED leads to a lower R_FeII compared to the other SED - the new SED predicts a R_FeII value which is lower by about -0.3. But, for the R_CaT, we retrieve an estimate that is only slightly higher than the previous estimate from P21, i.e. higher by about 0.02-0.04 that is well within the scatter in the observed flux ratio obtained by Persson (1988) and Marinello et al. (2016). Thus, we can conclude that there is not a significant change in the predicted R_FeII and R_CaT (and their corresponding line EWs) with the incorporation of a “better” SED, and hence, the conclusions obtained from our earlier analysis in P21 remain valid. In addition, we show the change in the line EWs in Figure 5, wherein we can notice that for the location in the $\log{U}-\log{n_{H}}$ plane with the optimal values for EWs, i.e. $\sim$40Å for Fe ii and $\sim$100Å for H$\beta$ (see Table A1 in Mart´ınez-Aldama et al., 2021b), which corresponds to a log U $\lesssim$ -3.0 around BLR densities log ${n_{H}}$ $\sim$ 10¹² cm^-3, we have a $\Delta$Fe ii around 10-20Å, while for H$\beta$ this difference rises to be around 30-50Å. This leads to the slump in the R_FeII that we notice in Figure 4. While in the case of the $\Delta$CaT, there is only a marginal change in the EW in the same region, i.e. about 5-10Å, which confirms the almost no change in the R_CaT plots. Thus, the EWs for these LILs provide a better insight to the changes arising due to the change in the SED - an increase in the prominence leads to an increase in the line EWs for these LILs, while the ratios (R_FeII and R_CaT) remain rather unaffected.

In this paper, we focus on the issue of the spectral energy distribution (SED) characteristic to a prototypical Type-1 Narrow-line Seyfert galaxy (NLS1) - I Zw 1. We extract the continuum from quasi-simultaneous spectroscopic measurements ranging from the near-UV ($\sim$1200Å) to the near infrared ($\sim$24000Å) to construct the SED and supplement it with archival X-ray measurements available for this source. Using the photoionization code CLOUDY, we assess and compare the contribution of the prominent “Big Blue Bump” seen in our SED versus the SED used in our previous work, wherein the latter was constructed from archival, multi-epoch photometric measurements. Following the prescription from our previous work, we constrain the physical parameter space to optimize the emission from these low-ionization lines (LILs) and discuss the implication of the use of a “better” SED. We find:

• •

  There is only a very slight difference in the estimated flux ratios, i.e. R_FeII and R_CaT when we replace the SED from the one used in our previous works (Panda et al., 2020; Panda, 2021) to a new, better one which is made by extracting the continuum from spectra for I Zw 1 ranging from the UV to the NIR.
  

• •

  The inclusion of the X-ray data in the construction of the new SED doesn’t lead to any significant change in the retrieved flux ratios (or their corresponding line EWs).
  

• •

  The BLR clouds need to be selectively overabundant in iron to reproduce the observed Fe ii emission, i.e. up to 3-10 times the solar values. On the contrary, the CaT emission predicted from our models agrees with the observed values.
  

• •

  The analysis presented here and in P21 highlight the importance to consider the comparison of the line EWs, in addition to just the flux ratios, which leads to a significant improvement to break the degeneracy and exclusion of imposter solutions. The EWs for these LILs provide a better insight to the changes arising due to the change in the SED - an increase in the prominence leads to an increase in the line EWs for these LILs, while the ratios (R_FeII and R_CaT) remain rather unaffected.
  

• •

  We are successful in constraining the physical parameter space to optimize the emission from these low-ionization lines originating from the BLR and re-affirm the similarity in the location leading to the emission of the Fe ii and CaT emission.

Our conclusions obtained suggest the importance of constructing better SEDs and utilizing them instead of generic ones. In this work, we focused on the study and analysis of the emission lines pertaining to the BLR which are governed by the ionizing photon flux especially from the energy range that corresponds to the radiation from the accretion disk and the Comptonized corona. We show that the continuum extraction from the broad-band spectrum for I Zw 1 is able to highlight the prominence of the Big Blue Bump feature in the SED. This was not accounted for in the earlier works and led to interesting results, especially in the line EWs recovery where we see a marked increase in the recovered EWs compared to our previous works. There is still progress needed to fully account for the X-ray radiation, i.e. having a self-consistent SED where we account for the continuum in the X-ray and construct an SED where we also model the region where we have the galactic absorption. This can be tested with currently available xspec models, e.g. OPTXAGN (Done et al., 2012). Also, there is a need to test our findings for other sources similar to I Zw 1. In addition to these, there is a need to test and account for the changes in the accretion disk structure (and the corona) as a function of the increasing accretion rate. This increase in the accretion rate has been shown to modify the standard Shakura and Sunyaev (1973) disk to a more intricate, slim disk (Abramowicz et al., 1988) solution. Such a change can lead to the development of a inner funnel in the very inner regions of the disk one that is the closest to the BH. This can modify significantly the overall photon energy distribution and make the radiation from these regions more anisotropic. BLR clouds which are located closer to the disk surface would preferably see a continuum that contain less energetic while the distant observer may see a very different SED, more rich in high energy photons coming from the region around the inner-most stable circular orbit of the BH. This anisotropy can help quantify the difference in the SED seen by the BLR clouds in comparison to the distant observer (see Panda, 2021, for a qualitative overview). This work is also in progress and will be presented in a forthcoming work.

CLOUDY v17.02 (Ferland et al. 2017); MATPLOTLIB (Hunter 2007); NUMPY (Oliphant 2015)