Broad-line region structure and line profile variations in the changing look AGN HE 1136-2304 ^††thanks: Based on observations obtained with the Southern African Large Telescope.

We present all final reduced optical spectra of HE 1136-2304 taken during the 2014/2015 variability campaign in Fig. 1. These results clearly show variations in the continuum intensities. The mean and rms spectra of HE 1136-2304 are shown in Fig. 2. The rms spectrum presents the variable part of the emission lines. The spectrum was scaled by a factor of 6.9 to allow for a better comparison with the mean spectrum and for enhancing weaker line structures. We note that all wavelengths referred to in this section are rest-frame wavelengths.

The integration limits of the broad emission lines and continuum regions are given at the bottom of the spectra. To select the continuum regions, we inspected the mean and rms spectra for regions that are free of both strong emission and absorption lines. The final wavelength ranges used for our continuum flux measurements are given in Table 2. A continuum region at 5100 Å is often used in studies of the variable continuum flux in AGN. Normally, this region is free of strong emission lines and close to the [O iii] $\lambda$5007 flux calibration line. However, in our case this region falls in the gap between the blue and central CCD chip. Therefore we set a nearby continuum range at 5360 Å. In addition to this continuum range, we determined the continuum intensities at three additional ranges (at 4570, 6235, and 6835 Å; see Fig. 2 and Tab. 2). We used these continuum regions for creating pseudo-continua below the variable broad emission lines. We neglected a possible contribution of FeII blends to the continuum flux based on our mean and rms spectra.

We integrated the broad Balmer and Helium emission-line intensities between the wavelength boundaries given in Tab. 2. Before integrating each emission line flux, we subtracted a linear pseudo-continuum defined by the boundaries given in Tab. 2 (Col. 3). We did not consider the H$\gamma$ line in our studies as it was not possible to determine a reliable continuum at the blue side of this line. The results of the continuum and line intensity measurements are given in Tab. 1. Additionally, we present in this table the flux values obtained for our first spectrum taken on 2014 July 7.

We present the light curves of the continuum fluxes at 4570 Å and 5360 Å as well as those of the integrated emission line fluxes of H$\alpha$, H$\beta$, He ii $\lambda 4686$, and He i $\lambda 5876$ in Fig. 3.

Some statistics of the emission line intensity and continuum variations are given in Tab. 4. We indicate the minimum and maximum fluxes F${}_{\text{min}}$ and F${}_{\text{max}}$, peak-to-peak amplitudes R${}_{\text{max}}$ = F${}_{\text{max}}$/F${}_{\text{min}}$, the mean flux over the period of observations $<$F$>$, standard deviation $\sigma_{F}$, and fractional variation F${}_{\text{var}}$ (see Paper 1).

We calculated the Balmer decrement H$\alpha$/H$\beta$ of the broad components after subtraction of the narrow components for the individual epochs. The results are shown in Fig. 4 as a function of the continuum intensity at 4570 Å. Fig. 5 gives the Balmer decrement as a function of the H$\beta$ line intensity. The Balmer decrement H$\alpha$/H$\beta$ of the broad components takes values between 3.5 and 7.5. On the other hand, we measure a constant value of 2.81 for the Balmer decrement H$\alpha$/H$\beta$ of the narrow components. There is a clear trend that the Balmer decrement H$\alpha$/H$\beta$ of the broad component decreases with increasing luminosity.

We determined normalized mean and rms profiles of the Balmer and Helium lines after subtracting the continuum fluxes in each individual spectrum using the continuum windows listed in Tab. 2. Figs. 7 to 14 show the mean and rms profiles of the Balmer lines H$\alpha$ and H$\beta$ as well as those of the He lines He i $\lambda 5876$ and He ii $\lambda 4686$ in velocity space. The rms profiles illustrate the line profile variations during our campaign. The constant narrow components disappear almost completely in these rms profiles. However, the mean profiles contain strong narrow line components in addition to their broad line components. We subtracted the narrow Balmer and Helium line components in all the mean profiles by subtracting a scaled [O iii] $\lambda$5007 line profile as a template. Furthermore, we subtracted the narrow [N ii] $\lambda 6584$ line in the mean H$\alpha$ profile and removed the [Fe iii] $\lambda 4861$ line in the mean He ii $\lambda 4686$ profile as well. After subtracting these narrow components we compared the mean and rms profiles of the broad emission lines with each other in a more accurate way.

We present the normalized mean profiles (with and without narrow components) and rms profiles for all four Balmer and Helium emission lines in Figs. 7 to 9.

A comparison of all normalized mean profiles including their narrow components is shown in Fig.10. We then present a comparison of the mean profiles after subtracting the narrow components; this comparison is shown in Fig.11 for the mean profiles of H$\alpha$ and H$\beta$. For getting a hint on line asymmetries their flipped profiles at $v=0$ km s^-1are depicted as well.

Fig.12 presents the mean Helium profiles after subtracting their narrow components. The profile of H$\beta$ has also been added for comparison purposes.

Fig.13 shows a comparison of the H$\alpha$ and H$\beta$ rms line profiles normalized with respect to their central component. Again their flipped profiles at $v=0$ km s^-1are depicted to highlight the additional strong red components and the weak blue component.

Finally, Fig.14 gives the rms profiles of the Helium lines. Again, the H$\beta$ line profile has been added for comparison.

We determined the linewidths (FWHM) of the mean and rms line profiles of all the Balmer and Helium lines. These measurements were performed with respect to the maximum of the central component and are listed in Tab. 5. In addition, we parameterized the linewidths of the rms profiles by their line dispersion $\sigma_{\text{line}}$ (rms widths) (Fromerth & Melia2000; Peterson et al.2004).

The Balmer line profiles (mean and rms) exhibit linewidths (FWHM) between 2500 and 4000 km s^-1. H$\alpha$ shows the narrowest broad line profiles. The Helium lines (He i $\lambda 5876$ and He ii $\lambda 4686$) are always broader than the Balmer lines by 500 to 1500 km s^-1. Furthermore, the outer blue wing is more extended in the higher ionized He lines in comparison to the Balmer lines (Figs.13, 14). This might be an indication of an additional outflowing component in the inner broad-line region (BLR).

An additional strong red component emerges in the rms and mean profiles of the H$\alpha$ and H$\beta$ lines at 1380 km s^-1and varies with a larger amplitude than the rest of the line profile. Furthermore, this component shows a relatively stronger variation in H$\beta$ in comparison with H$\alpha$ (see Fig.13) by a factor of 1.5. An indication of the existence of this red component is present in the He ii $\lambda 4686$ line as well (Fig.14). No clear indication of this component is visible in the red wing of the He i $\lambda 5876$ line as that region coincides with the NaD absorption. An additional blue component, nearly symmetrical to the red component, appears in the rms profile of H$\beta$ at around -1400 km s^-1. (Fig.13). However, this blue component is by far weaker than the red component. The existence of blue and red components in the line profiles – in addition to the central component – are an indication that the broad line-emitting region is connected with an accretion disk structure.

The masses of the central black holes in AGN can be estimated from the width of the broad emission line profiles, based on the assumption that the gas dynamics are dominated by the central massive object, by evaluating $M=f\,c\,\tau_{\text{cent}}\,\Delta\,v^{2}\,G^{-1}$. It is necessary to know the distance of the line-emitting region. Characteristic distances of the individual line-emitting regions are given by the centroid $\tau_{\text{cent}}$ of the individual cross-correlation functions of the emission-line variations relative to the continuum variations (e.g., Koratkar & Gaskell1991; Kollatschny & Dietrich1997). The characteristic velocity $\Delta v$ of the emission-line regions can be estimated from the FWHM of the rms profiles or from the line dispersions $\sigma_{\text{line}}$.

The scaling factor $f$ in the equation above is on the order of unity and depends on the kinematics, structure, and orientation of the BLR. This scaling factor may differ from galaxy to galaxy, for example, depending on whether we see the central accretion disk including the BLR from the edge or face-on. We compared the central black hole mass value of HE 1136-2304 with values of the black hole masses derived for other AGN and adopt a mean value of $f=5.5$ (e.g., Onken et al.2004; Grier et al.2012). This f-value might be too high by a factor of two when comparing the black hole masses with inactive galaxies (Graham et al.2011). Nevertheless, using $f=5.5$, we calculate a mean black hole mass (see Tab. 8, col. 3) of

based on the derived delays of the integrated Balmer and Helium lines (see Tab. 6) and on the line dispersions $\sigma_{\text{line}}$ (see Tab. 5). All BH masses based on the individual lines agree with each other within the error limits.

Using the linewidths FWHM (rms) (see Tab. 5) we calculated a mean black hole mass of $M=11.4\pm 9.3\times 10^{7}M_{\odot}$ (Tab. 8, col. 4). However, in that case we did not correct for the contribution of turbulent motions to the width of the line profiles; this is covered in section 3.7. After correcting the emission linewidths (FWHM) for their contribution of turbulent motions (Tab. 9, col. 4), we derive a mean black hole mass of

(Tab. 8, col. 2). Again, all individual BH masses based on this method agree with each other within the error limits.

We first calculated the cross-correlation lags of the integrated Balmer and helium lines with respect to the continuum as mentioned in section 3.3. Now we investigate the profile variations of these lines in more detail by calculating the lags of individual line segments. The way we proceed has been described before in our studies of line profile variations in Mrk 110 (Kollatschny & Bischoff2002; Kollatschny2003), Mrk 926 (Kollatschny & Zetzlkollatschny10 2010), and 3C120 (Kollatschny et al.2014).

We sliced the velocity profiles of the Balmer and Helium lines into velocity segments with a width of $\Delta$$v=400$ km s^-1, which corresponds to the spectral resolution of our spectra. A central line segment was integrated in the velocity range -200 km s^-1$\leq v\leq$ 200 km s^-1. Afterward, we measured the intensities of all subsequent velocity segments from $v=-9\,800$ to $+9\,800$ km s^-1and compiled their light curves. Light curves of the central Balmer and Helium line segments and selected blue and red segments at $800$, $2000$, and $4000$ km s^-1 are shown in Figs. 30 to 33. For comparison, the light curve of the continuum flux at 4570 Å is given in all these figures as well. We computed the maximal correlation coefficient and time delay $\tau_{\text{cent}}$ of all line segment ($\Delta{}v=400$ km s^-1) light curves of the Balmer and Helium lines with respect to the 4570 Å continuum flux light curve. The derived time delays of the segments are shown in Figs. 21 to 23 as functions of distance to the line center (blue scale). The white lines in Figs. 21 to 23 delineate the contour lines of the correlation coefficient at different levels. The green line shows the line profile of the mean spectrum for comparison.

The following statements can be made based on these figures. There is a general trend that the Helium line response is faster than the response of the Balmer lines, as already known from the integrated lines. The velocity-delay maps are very symmetric with respect to the line center. The light curves of the emission line centers are delayed by 10 to 20 days with respect to the continuum variations, while the outer line wings at distances of $+/-$2000 to $+/-$3000 km s^-1 respond much faster to continuum variations and only show a delay of 0 to 10 days. The delay in the outer line wings at distances of $+/-$4000 km s^-1 is even negative with respect to the optical continuum. It has been discussed in Zetzl et al.(2018) that the observed optical continuum is delayed with respect to the ionizing continuum in the UV and X-ray bands. The outer blue wing of the H$\beta$ line shortward of $-6000$ km s^-1) is blended with the red wing of the HeII$\lambda$4686 line (see Fig.2 as well).

Information about the BLR structure in Seyfert 1 galaxies can be derived from the profiles of the broad emission lines together with variability studies (Kollatschny & Zetzlkollatschny11 2011, 2013a, 2013b, 2013c). The broad emission line profiles can be parameterized by the ratio of their full-width at half maximum (FWHM) to their line dispersion $\sigma_{\mathrm{line}}$. We were able to show that there exists a general relation between the FWHM and the linewidth ratio FWHM/$\sigma_{\mathrm{line}}$. The linewidth FWHM primarily reflects the line broadening of the intrinsic Lorentzian profiles due to rotational motions of the broad line gas. The intrinsic Lorentzian profiles themselves are associated with turbulent motion (see also Goad et al.goad12 2012) and different emission lines turn out to exhibit, on average, characteristic turbulent velocities within a narrow range (Kollatschny & Zetzlkollatschny11 2011).

We determined the rotational velocities and turbulent velocities that belong to the individual line-emitting regions in the same way as we have done it before for other Seyfert galaxies (Kollatschny & Zetzlkollatschny11 2011, 2013a, 2013b, 2013c): Based on the observed linewidths (FWHM) and linewidth ratios FWHM/$\sigma_{\mathrm{line}}$ we determined the locations of the individual lines in Fig. 24.

In this figure, the grid, resulting from model calculations, presents theoretical linewidth ratios based on Lorentzian profiles that are broadened owing to rotation. We read the widths of the Lorentzian profiles and the rotational velocities of the individual lines from their positions between the contour lines of constant Lorentzian linewidth and the vertical contour lines representing different $v_{\text{rot}}$. The FWHM/$\sigma_{\mathrm{line}}$ versus FWHM grid based on our model calculations is publicly available.²²2http://www.astro.physik.uni-goettingen.de/~zetzl/blrvelo/ The FWHM and FWHM/ $\sigma_{\mathrm{line}}$ values we obtained for HE 1136-2304 are given in Tab. 9 together with the derived $v_{\text{turb}}$ and $v_{\text{rot}}$ velocities of the Balmer and Helium lines. It has been shown that the region of each emission line has a characteristic mean turbulent velocity within a narrow range (Kollatschny & Zetzlkollatschny11 2011, 2013a; 2014). We derived the following mean turbulent velocities belonging to the emitting regions of the individual lines: 400 km s^-1for H$\beta$, 700 km s^-1for H$\alpha$, 800 km s^-1for He i $\lambda 5876$, and 900 km s^-1for He ii $\lambda 4686$.

In the next step, we determine the heights of the line-emitting regions above the midplane as we have before for other Seyfert galaxies. The ratio of the turbulent velocity $v_{\text{turb}}$ with respect to the rotational velocity $v_{\text{rot}}$ in the line-emitting region gives us information on the ratio of the height $H$ with respect to the radius $R$ of the line-emitting regions as presented in Kollatschny & Zetzl kollatschny11 (2011, 2013a), i.e.,

The unknown viscosity parameter $\alpha$ is assumed to be constant and to have values of 0.1 to 1 (e.g., Frank et al.2003). For simplicity we assume a value of 1 in the present investigation. The distance of the line-emitting regions of the individual lines is known from reverberation mapping (section 3.3). We present in Tab. 9 the derived heights above the midplane of the line-emitting regions in HE 1136-2304 (in units of light days) and the ratio $H/R$ for the individual emission lines.

The BLR structure of HE 1136-2304 is shown in Fig. 25 as a function of distance to the center and height above the midplane. The dot at radius zero gives the size of the Schwarzschild radius $R_{\mathrm{S}}=4.31\times 10^{-3}\,\mathrm{ld}=1.1\times 10^{13}\,\mathrm{cm}$ for a black hole mass (with $M=3.8.\times 10^{7}M_{\sun}$) multiplied by a factor of ten. The label on top of the figure gives the distances of the line-emitting regions in units of the Schwarzschild radius.

As has been observed previously in other galaxies, the He ii $\lambda 4686$ line originates at the shortest distance from the center and the smallest height above the midplane in comparison to the Balmer and HeI lines. In comparison to H$\beta$, H$\alpha$ originates at a larger distance from the midplane.

We thoroughly investigated the spectroscopic variability behavior of HE 1136-2304 by taking 16 spectra over a period of six months between February to August 2015. The fractional variability F${}_{\text{var}}$ was on the order of 0.1 in the optical continuum without correcting for the host galaxy flux. After correcting for the host galaxy contribution, the fractional variability F${}_{\text{var}}$ of the continuum amounted to $0.25-0.3$ (see Paper I). The integrated Balmer and Helium lines showed F${}_{\text{var}}$ values of 0.1 to 0.5 and the higher ionized lines originating closer to the center varied with stronger amplitudes. These results describing the continuum and emission line variability are similar to those detected in other variable Seyfert galaxies such as NGC 5548 (Peterson et al.2004), Mrk 110 (Kollatschny et al.2001), or 3C 120 (Kollatschny et al.2014). This confirms that the variability behavior of this changing look AGN is similar to that of other Seyfert galaxies.

The Balmer decrement H$\alpha$/H$\beta$ of the narrow components has a value of 2.81. This corresponds exactly to the expected theoretical line ratio (Case B) without any reddening. However, the Balmer decrement H$\alpha$/H$\beta$ of the broad components varies with the continuum and/or Balmer line intensity. For example, the broad line Seyfert galaxy NGC 7693 showed the same behavior based on long-term variability studies over a period of 20 years (Kollatschny et al.2000): the Balmer decrement decreased with increasing H$\beta$ flux.

Heard & Gaskellheard16 (2016) proposed a model with additional dust reddening clouds interior to the narrow-line region causing higher Balmer decrements in the BLR. In contrast to this model, there might be important optical depth effects in the BLR itself explaining the observations. This is consistent with the finding that H$\alpha$ originates at twice the distance of H$\beta$. A similar radial stratification as seen in HE 1136-2304 has been observed in, for example, Arp 151 (Bentz et al.2010) as well. It has been discussed by Korista and Goad(2004) that the radial stratification is a result of optical-depth effects of the Balmer lines: the broad-line Balmer decrement decreases in high continuum states and steepens in low states exactly as observed in HE 1136-2304 (see Fig. 4). The continuum varied by a factor of nearly two during our campaign in 2015. However, we did not detect simultaneous variations of the Seyfert subtype during our observing period of seven months. A variation of Seyfert subtypes might be connected with stronger continuum amplitudes and/or longer timescales as has been seen before, for example, in Fairall 9 (Kollatschny et al.1985).

Now we want to test whether HE 1136-2304 follows the general trend in the radius-luminosity relationship for AGN (Kaspi et al.2000; Bentz et al.2013). We determined a continuum luminosity $\log\lambda\text{L}_{\lambda}$ of 42.6054 erg s^-1 ($0.47\times 10^{-15}$ erg s^-1 cm^-2 Å^-1) in the optical at 5100 Å after correction for the contribution of the host galaxy (Zetzl et al.2018). Furthermore, we derived a mean radius of 7.5 light days for the H$\beta$ line-emitting region based on the delay of the integrated H$\beta$ line variability curve with respect to the optical continuum light curve.

Fig. 26 shows the optical continuum luminosity and H$\beta$-optical lags for HE 1136-2304 (red), for NGC 5548 based on different variability campaigns (black; Pei et al.2017 based on Kilerci-Eser et al.2015 and Denney et al.2009), and a sample of other AGN excluding NGC 5548 (green; Bentz et al.2013). The black line is the linear least-squares fit to the NGC 5548 data as presented by Pei et al.(2017). The $R-L$(5100Å) relationship is given by

where K is the origin and $\beta$ is the slope. The red solid line gives the best-fit linear regression to the whole AGN sample. The data of HE 1136-2304 is in very good accordance with the general H$\beta$ lag versus the optical continuum luminosity relation.

The red solid line has a slope $\beta$ of 0.53, which is therefore identical to the best-fit slope of $0.533^{+0.035}_{-0.033}$ of Bentz et al.(2013). This value is very close to the value of 0.5 expected from simple photoionization arguments, i.e.,

(e.g., Kaspi et al.2000; Bentz et al.2013 and references therein).

We tested whether the $\beta$ slope approaches values even closer to 0.5 or whether the Pearson correlation coefficient becomes higher if we add a few light days to the H$\beta$ radius. Such an additional delay might be caused by the fact that the optical continuum is generally delayed by a few light days with respect to the driving X-ray light curve (Zetzl et al.2018, Shappee et al.2014; Fausnaugh et al.2016).

We added additional lags of one to eight light days to all H$\beta$-optical lags to take into account a systematic delay of the optical bands with respect to the driving X-ray flux. A time delay of eight light days is an upper limit based on the correlation of the optical band light curves with respect to the XRT light curve (Zetzl et al.2018). Figs. 27 and 28 show the H$\beta$ lag versus optical continuum luminosity diagrams with an additional lag of one and four days, respectively, taking into account the optical-X-ray lag.

Tab. 10 gives the Pearson correlation coefficient for the relation between optical continuum luminosities and H$\beta$-optical lags. The H$\beta$ lags have been modified assuming additional lags (in units of days) for the optical lag with respect to the driving X-ray source. Furthermore, we present the $\beta$ values for the additional delays that have been assumed. We get the highest correlation coefficient for an additional delay of four days. We reached a $\beta$ slope of exactly 0.5 for an additional delay of one day.

The higher ionized broad emission lines originate at smaller radii as shown in section 3.3. Furthermore, the integrated H$\alpha$ originates at a distance of 15 light days and therefore at twice the distance of H$\beta$ (see Tab. 6). Moreover, it has been shown that the higher ionized lines originate closer to the midplane of the accretion disk in comparison to the lower ionized lines. We presented the BLR structure as a function of distance to the center and height above the midplane (Fig. 25). The He ii $\lambda 4686$ line originates closest to the midplane. H$\alpha$ originates at a larger distance from the midplane in comparison to H$\beta$. Such a trend has been observed before in other galaxies as NGC 7469 (Kollatschny & Zetzlkollatschny13c 2013c) and 3C 120 (Kollatschny et al.2014).

A second trend has been found when comparing the H$\beta$ distances above the midplane for different active galaxies: galaxies showing the broadest H$\beta$ linewidths originate closest to the midplane, while galaxies showing the narrowest H$\beta$ linewidths originate at the largest distance to the midplane (Kollatschny et al.2014). The linewidths (with respect to the individual lines) are therefore a characteristic for the height of the line-emitting regions above the midplane. We present the height-to-radius ratio and FWHM of H$\beta$ for a sample of AGN (Kollatschny et al.2014) and for HE 1136-2304 in Tab. 11.

The height-to-radius ratio for H$\beta$ is largest for galaxies showing narrow emission lines and smallest for galaxies with broad lines. The overall picture we derived for the BLR region structure previously in Kollatschny & Zetzlkollatschny13c (2013c) and Kollatschny et al.2014 is confirmed by the additional emission line data of HE 1136-2304. The derived height-to-radius ratio for HE 1136-2304 confirms the general trend (see Fig. 29). Again, the HE 1136-2304 data support the picture that the broad emission line geometries of AGN are not simply scaled-up versions depending only on the central luminosity (and central black hole mass).