Catalina Sky Survey Continued to Check in on J1011 5442

Abstract

We report the discovery of a new 'changing-look' quasar, SDSS J101152.98+544206.4, through repeat spectroscopy from the Time Domain Spectroscopic Survey. This is an addition to a small but growing set of quasars whose blue continua and broad optical emission lines have been observed to decline by a large factor on a time-scale of approximately a decade. The 5100 Å monochromatic continuum luminosity of this quasar drops by a factor of >9.8 in a rest-frame time interval of <9.7 yr, while the broad Hα luminosity drops by a factor of 55 in the same amount of time. The width of the broad Hα line increases in the dim state such that the black hole mass derived from the appropriate single-epoch scaling relation agrees between the two epochs within a factor of 3. The fluxes of the narrow emission lines do not appear to change between epochs. The light curve obtained by the Catalina Sky Survey suggests that the transition occurs within a rest-frame time interval of approximately 500 d. We examine three possible mechanisms for this transition suggested in the recent literature. An abrupt change in the reddening towards the central engine is disfavoured by the substantial difference between the time-scale to obscure the central engine and the observed time-scale of the transition. A decaying tidal disruption flare is consistent with the decay rate of the light curve but not with the prolonged bright state preceding the decay; nor can this scenario provide the power required by the luminosities of the emission lines. An abrupt drop in the accretion rate on to the supermassive black hole appears to be the most plausible explanation for the rapid dimming.

1 INTRODUCTION

The dichotomy between Type 1 and Type 2 active galactic nuclei (AGN) is among the most studied in astrophysics (for a review, see Netzer 2015). The classes are empirically determined; Type 1 AGN are characterized by having both broad (of the order of 10³ km s⁻¹) as well as narrow (of the order of 10² km s⁻¹) emission lines, whereas only the narrow lines are present in the spectra of Type 2 AGN. The intermediate Type 1.8 and 1.9 classifications describe objects with weak or absent broad Hβ emission, respectively, but a broad Hα line (Osterbrock 1981). In the simplest scenario, these objects can all be unified by orientation (Antonucci 1993); Type 2 AGN are viewed edge-on, with an equatorial dusty structure obscuring the line of sight down to the broad-line region (BLR), while Type 1 objects have an unobscured, face-on view. Early support for this scheme came from ground-breaking views of broad emission lines in Type 2 objects observed in scattered light (Antonucci & Miller 1985). In this picture, intermediate types are explained by partial obscuration or reddening by optically thin dust (Stern & Laor 2012). In all cases, the narrow emission lines originate farther from the central engine in the narrow-line region (NLR), and are not hidden.

While the simplicity of the above orientation-based scheme is compelling, many have suggested that it does not fully explain the Type 1–Type 2 dichotomy. An alternative possibility is that the link between classifications might be evolutionary (e.g., Penston & Perez 1984). Based on NGC 4151 (see also Lyutyj, Oknyanskij & Chuvaev 1984) and 3C 390.3, which were both initially observed to have broad emission lines that later disappeared, Type 2 AGNs may be Type 1s that have 'turned off'. Variations along this line invoke age, environment, or luminosity effects and are summarized by Gaskell (2014) (see also, Villarroel & Korn 2014).

Objects that transition between Types 1 and 2 provide a unique opportunity to study the evolutionary side of unification. Although not common, the number of known objects in this class has been growing. The term 'changing-look' was initially used to describe X-ray AGN with variable obscuration (e.g. Matt, Guainazzi & Maiolino 2003; Puccetti et al. 2007; Bianchi et al. 2009; Risaliti et al. 2009; Marchese et al. 2012), but has lately widened to include objects that transition between Types 1 and 2 (an optical classification). There are a handful of objects like NGC 4151 and 3C 390.3, where the broad emission lines and quasar continuum fade or disappear completely below the detection limit (Collin-Souffrin, Alloin & Andrillat 1973; Tohline & Osterbrock 1976; Denney et al. 2014; Sanmartim, Storchi-Bergmann & Brotherton 2014; Barth et al. 2015) as well as instances where the transition goes in the other direction and the AGN 'turns on' (Cohen et al. 1986; Storchi-Bergmann, Baldwin & Wilson 1993; Aretxaga et al. 1999; Eracleous & Halpern 2001; Shappee et al. 2014). The time-scales for the transition are typically found to be of the order of years, and obscuration or changes in accretion rate are often the preferred explanation. However, these objects have all been at notably low redshift and luminosity.

LaMassa et al. (2015) report the discovery of the first changing-look quasar, the most luminous and distant observed to date, with a redshift of z = 0.31. SDSS J015957.64+003310.5 (hereafter J0159+0033), is an X-ray-selected quasar from the Stripe 82X survey (LaMassa et al. 2013a,b) that transitioned from a Type 1 to a Type 1.9 within a ∼9 yr period. Initially observed in the first data release of the Sloan Digital Sky Survey (SDSS DR1; Abazajian et al. 2003), J0159+0033 appears as a typical, Type 1 broad-line quasar. In 2010, it was re-observed in the Baryon Oscillation Spectroscopic Survey (BOSS; Dawson et al. 2013) of the Sloan Digital Sky Survey - III (SDSS-III; Eisenstein et al. 2011), revealing a galaxy spectrum with only a weak quasar continuum and residual broad Hα. Archival X-ray observations of its bright and dim states show that the dimming in the X-ray flux is consistent with the optical continuum, with no strong evidence for obscuration in either band. Although variable obscuration cannot be emphatically ruled out, LaMassa et al. (2015) favour an interpretation where the changes in J0159+0033 are due to a decrease in accretion power.

The discovery of luminous analogues of the nearby changing-look AGNs is significant because it demonstrates that such abrupt transitions can occur in quasars where the black holes are more massive and the physical time-scales correspondingly longer than their low-redshift counterparts. Motivated by the above considerations, Ruan et al. (2015) performed an archival search of the SDSS DR12 spectroscopic data base, in an effort to find more such quasars. Their criteria target objects at all redshifts that exhibit significant changes in repeat SDSS observations, either transitioning from quasar to galaxy or vice versa. In addition to J0159+0033, which is recovered, they identify two new changing-look quasars. In both cases, the changes in the spectrum are best explained by changes in the luminosity of a pre-existing accretion flow rather than tidal disruption events (TDEs), primarily because the observed narrow-line ratios are indicative of long-term AGN activity. Thus, it appears that these are examples of a previously unappreciated mode of quasar variability: prolonged 'on-' and 'off-states'. This type of variability may correspond to the bursting behaviour found simulations, such as those of Novak, Ostriker & Ciotti (2011), where the accretion rate on to the black hole is intermittent because of the interaction of the quasar outflow with the gas supplied by stellar processes (see also the related discussion by Schawinski et al. 2015; Merloni et al. 2015). Alternatively, the variability may be the result of thermal instabilities that drive the accretion disc to transition between different physical states (Ichimaru 1977). Such thermal instabilities may follow from small changes in the accretion rate, but the resulting large, abrupt change in the luminosity in this case would come from the transition of the disc state, e.g. to an advection-dominated flow (Narayan & Yi 1994), rather than a sudden cessation in accretion and draining of the disc.

In this work, we report the discovery of a new changing-look quasar, the first found via visual inspection of the Time-Domain Spectroscopic Survey (TDSS; Morganson et al. 2015). The source, SDSS J101152.98+544206.4 (hereafter J1011+5442), is a z = 0.246 quasar initially observed for spectroscopy by the SDSS in early 2003 that appeared in the third, fifth, and seventh SDSS quasar catalogues (Schneider et al. 2005, 2007, 2010). It was targeted again by TDSS in the Few Epoch Spectroscopy (FES) programme to study variability of the broad Balmer emission lines in a sample of normal, bright, low-redshift quasars, and a second spectrum was obtained in early 2015. The main goals of this work are to describe the behaviour of J1011+5442 and set the stage for a systematic search of TDSS for a sample of changing-look quasars. After collecting existing data for this object (Section2), we evaluate several physical scenarios that might explain the changes in J1011+5442 (Section3). We determine that the changes in spectral state are driven by changes in the accretion power, ruling out variable obscuration and TDEs. These results are summarized and discussed in the context of previous work in Section4. Throughout this work, we adopt the same cosmology used by LaMassa et al. (2015) of H ₀ = 70 km s⁻¹ Mpc⁻¹, Ω_Λ = 0.73, and Ω_m = 0.27, which leads to a luminosity distance of D _L  = 1.244 Gpc and a distance modulus of μ = 40.47 mag for J1011+5442.

2 DATA AND MEASUREMENTS

2.1 New and archival data

For this work, we collect the existing photometry and spectroscopy of J1011+5442. The bright-state spectrum was taken on 2003 January 13 through a 3.0 arcsec diameter fibre and covers the range 3800–9200 Å with a resolution of R ∼ 2000 (York et al. 2000; Smee et al. 2013) and was calibrated with the v5_3_12 version of the spectroscopic pipeline. Imaging and photometry are also available from the SDSS, taken almost a year earlier on 2002 February 13. In Fig.1, we show the SDSS colour image of J1011+5442 with the TDSS and SDSS fibre sizes and placements overlaid. This image shows that the host galaxy is extended with a visually asymmetric, potentially disturbed, morphology. Because the source appears extended, we adopt the Petrosian magnitudes, which use a large aperture (a 3.6 arcsec diameter in the r band, corresponding to 21 kpc) to capture virtually all of the light. These ugriz magnitudes are 18.32 ± 0.01, 18.20 ± 0.01, 17.90 ± 0.08, 17.64 ± 0.09, and 17.52 ± 0.03. The corresponding absolute magnitude in the i band is M _i  = −22.82, which makes this object a quasar according to the definition of the SDSS DR10 quasar catalogue (Schneider et al. 2010). From the ugriz colours and the transformations of Jester et al. (2005) appropriate for quasars of z < 2.1, we obtain V = 18.01 ± 0.01 and R = 17.64 ± 0.01. The object was re-observed during the TDSS, a subprogramme of the Extended Baryon Oscillation Spectroscopic Survey (eBOSS), on 2015 February 20. Specifically, J1011+5442 was targeted by an FES programme to study the variability of the Balmer lines in high signal-to-noise ratio quasar spectra. The TDSS spectroscopic fibre had a diameter 2.0 arcsec and was centred on a position within 0.1 arcsec of the SDSS fibre position. The new TDSS spectrum covers the range 3600–10 000 Å at R ∼ 2000 and was calibrated with the v5_7_8 version of the spectroscopic pipeline. We apply two corrections to the spectra. To correct for dust extinction in the Milky Way, we deredden the spectra using a Seaton extinction law (Seaton 1979) with E(B −V) = 0.01 taken from the maps of Schlegel, Finkbeiner & Davis (1998). We then shift the spectrum to the rest frame using the SDSS pipeline redshift of z = 0.246.

Figure 1.

The SDSS explorer image of the host galaxy showing the fibre placement and size for TDSS (solid) and SDSS (dashed). The scale bar subtends 2.0 arcsec and North and East are up and left, respectively. The fibres are placed on the image based on the PLUG coordinates from the fits headers of the spectra and sized with 2.0 arcsec and 3.0 arcsec diameters for TDSS and SDSS, respectively.

In addition to the SDSS photometry in the dim state, there are infrared data available from the Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010) ALLWISE multi-epoch photometry catalogue. J1011+5442 was observed to have magnitudes of W1 = 13.99 ± 0.03, W2 = 12.86 ± 0.06, W3 = 9.7 ± 0.1, and W4 = 7.2 ± 0.2 on 2010 April 28 and W1 = 14.19 ± 0.09 and W2 = 13.08 ± 0.08 on 2010 November 5, where we list standard deviation of observations within days of each other as the magnitude uncertainty. These data do not appear to offer any additional information on the variability properties of this object. J1011+5442 has not been observed by the Chandra X-ray Observatory or XMM–Newton and it was not detected in the ROSAT all-sky survey (Voges et al. 1999).

J1011+5442 was observed regularly between 2002 December and 2008 June as part of the MIT Lincoln Laboratory Lincoln Near-Earth Asteroid Research (LINEAR) survey (Stokes et al. 2000). This survey is performed using a single, broad filter, so we adopt the transformation of Sesar et al. (2011) to recalibrate the LINEAR photometry:

\begin{eqnarray} g &=& g_{{\rm LINEAR}}-0.0574-0.004\,(g-i) \nonumber \\ &&+\,0.056\,(g-i)^2-0.052\,(g-i)^3+0.0262\,(g-i)^4. \end{eqnarray}

(1)

To apply the above equation, we use the SDSS g −i colour and convert the light curve to the V band using the SDSS g magnitude and the Jester et al. (2005) transformations. The LINEAR light curve is shown in Fig.2 and discussed in more detail below.

Figure 2.

Top: the light curve of J1011+5442, including photometry from the CSS (filled red circles), LINEAR (filled blue triangles), and the SDSS (open black square). The V magnitude inferred from the TDSS spectrum, observed on MJD 57073, is 19.350 ± 0.002 and is not plotted here. This value is almost certainly a lower limit on the true brightness of the AGN and host, as discussed in Section2. The horizontal bar represents the likely dimming time-scale, ∼500 d in the rest frame. Bottom: the median light curve, where each point represents the median of the observations in one year bins and the error bars represent the rms scatter in each bin. The dashed lines overlapping both panels indicate the timing of the SDSS and TDSS spectral observations.

Overlapping the LINEAR coverage, J1011+5442 was also observed between 2006 January and 2013 May as part of the Catalina Sky Survey (CSS; Drake et al. 2009). Although the observations for this survey are made in white (i.e. unfiltered) light, the light curves are reported in V-band magnitudes. We further refine the reported magnitudes using the listed transformation as a function of the V −R colour:

\begin{equation} V = V_{{\rm CSS}} +0.91\,(V-R)^{2}+0.04. \end{equation}

(2)

This transformation was selected from the three available because at the redshift of this quasar neither filter includes the strongest emission line, Hα, while both filters include the second strongest emission line, Hβ, in the spectral range where they overlap. Therefore, the V −R colour suffers the least from emission-line contamination, compared to other colours we could have used.

The total light curve, including the LINEAR and CSS photometry, for J1011+5442 is shown in Fig.2. This figure also includes the SDSS magnitude, which matches the LINEAR magnitudes and early-time measurements in the CSS light curve fairly well. The late-time TDSS fibre magnitude is V = 19.350 ± 0.002, but this corresponds to a 2.0 arcsec aperture, which excludes some of the light from the extended emission from the host galaxy. Therefore, we regard this value as a lower limit on the brightness of the AGN and host and do not include it in Fig.2. It is not unusual to see variability in a quasar light curve; variability of the order of tenths of a magnitude is typical for quasars on time-scales of a year (e.g. MacLeod et al. 2012). However, the systematic decline of the light curve by half of a magnitude, as displayed in Fig.2, over a time-scale shorter than several years is not typical behaviour.

From the light curve, we can estimate the time-scale for the quasar flux to decay. During the first and last years of observation, the average brightness of the quasar appears constant. We estimate the transition time-scale by visually determining where the light curve departs from the bright state and arrives at the dim state. The poor sampling and noise in the light curve make it difficult to determine exactly when the dimming occurs, but our best estimate is ∼500 d in the quasar rest frame. The relevant portion of the light curve is marked in the figure by a horizontal bar.

2.2 Spectral decomposition

We decompose the spectra interactively using the iraf task specfit (Kriss 1994), which fits multicomponent models to spectra by minimizing the χ² statistic. Components for the fit include a power law of the form f _λ =Aλ^−α to characterize the quasar continuum, Gaussians to characterize the emission lines, optical (Boroson & Green 1992) and UV (Vestergaard & Wilkes 2001) Feii templates that have been combined into a single template that covers our full wavelength range, and Charlot & Bruzual starlight templates (see Cales et al. 2013). The spectral decomposition is performed over the entire available wavelength range (3100–8000 Å in the dim state and 3100–7500 Å in the bright state) and all components are fitted simultaneously.

First, we decompose the dim-state spectrum because this spectrum is simpler than the bright-state spectrum. The starlight and narrow-line properties inferred from this spectrum aid us in the decomposition of the bright-state spectrum, as we explain below. We impose the following constraints on Gaussians representing narrow lines: within a doublet, the two emission lines are constrained to have identical velocity widths, a flux ratio of 3 when appropriate, and a constant wavelength ratio. We further group the lines by ionization and critical density and constrain the velocity widths to be the same across a group; Hα and Hβ form the first group, [Oii] λλ3726, 3729, [Nii] λλ6548, 6583, and [Sii] λλ6716, 6731 the second, and [Neiii] λλ3869, 3967 and [Oiii] λλ4959, 5007 the last. Finally, we also tie the wavelength of the narrow Hβ to that of the narrow Hα, because the Hα line is clear in the spectrum and the Hβ line from the quasar is weak enough that it is severely contaminated by the stellar lines. In order to achieve a satisfactory fit, this spectrum requires two stellar populations: one where the age is fixed at 10 Gyr and the normalization is a free parameter, plus a second where both the age and normalization are allowed to vary. This approach is similar to Canalizo & Stockton (2013), who find significant contributions of intermediate plus old stellar populations in quasar host galaxies. The broad lines (including the Feii) in this spectrum are very weak; only a symmetric broad Hα profile is visible and it is well characterized by a single Gaussian. The resulting fits are shown in Figs3 and4.

Figure 3.

The spectral fits for the bright-state (MJD 52652) and dim-state (MJD 57073) spectra. The data are shown in black and the best-fitting total model is shown in red. See Section2 for a description of the spectral decomposition and Fig.4 to see the flux uncertainties.

Figure 4.

The spectral decomposition in three wavelength windows including Hα (top), Hβ (middle), and strong stellar absorption lines (bottom). The data (black), uncertainties (grey), best-fitting model (red), old and young stellar templates (yellow, orange), power-law continuum (green), and emission lines (blue, see Table1 for a list) are displayed for the bright state (MJD 52652) on the left and the dim state (MJD 57073) on the right.

In the bright state, we fit the spectrum with all the components used in the dim state, plus the Feii template and a suite of Gaussians to characterize the broad lines. The broad Hα and Hβ lines are represented by two Gaussians each in order to describe their asymmetries, while all other broad lines are represented by single Gaussians. Since the size of the NLR (R _NLR ∼ 2.5 kpc based on the dim-state [Oiii] luminosity and the correlation of Hainline et al. 2013) should fall within the TDSS fibre, we fix all narrow-line properties to the best-fitting values from the dim-state decomposition. For the starlight templates, we fix the ages to the best-fitting values from the dim state but allow the normalizations to vary. Although we do not expect intrinsic variability from the host galaxy light, the size of the fibres used to obtain the observations were 3.0 arcsec and 2.0 arcsec in the bright and dim states, respectively, which creates the potential for differences between the spectra. Visual inspection of the SDSS image of this object reveals that the host galaxy extends beyond the diameter of the fibres, suggesting that the larger fibre may include more host-galaxy light. Thus, we do not require that the normalization of the stellar templates be identical in the two states.

The uncertainties in all parameters resulting from the spectral decomposition are determined via a full Monte Carlo simulation of the fitting procedure. We resample each spectrum by selecting for each pixel a new value for the flux density, drawn from a Gaussian distribution that has a mean equal to the measured flux density in that pixel, and a standard deviation equal to the measured uncertainty in the flux density. The best-fitting parameters for the observed spectrum are used as an initial guess, and the synthetic spectrum is refitted. This process is repeated 10³ times each for the bright- and dim-state spectra to generate distributions of all spectral fit parameters. The standard deviations of these distributions are taken to be the 1σ uncertainties in the measured properties.

The complete model fits to the bright- and dim-state spectra are shown in Fig.3. In Fig.4, we show an expanded view of three spectral windows of particular interest. In the latter figure, we also illustrate the individual model components. The bright- and dim-state emission-line measurements are listed in Table1; the continuum properties we describe here. In the bright state, the best-fitting power law has a normalization of (1410 ± 60) × 10⁻¹⁷ erg s⁻¹ cm⁻² Å⁻¹ at 1000 Å and an exponent of 3.42 ± 0.06. In the dim state, the best-fitting normalization and exponent are (11 ± 2) × 10⁻¹⁷ erg s⁻¹ cm⁻² Å⁻¹ and 1.9 ± 0.3, respectively. The full width at half-maximum (FWHM) of the lines in the Feii complex, which is only present in the bright state, is 1500 ± 200 km s⁻¹. The age of the young stellar template is 225 ± 4 Myrs in both states. The flux ratio of the young to old stellar templates at 5500 Å is 0.21 ± 0.01 in the bright state and 0.76 ± 0.01 in the dim state. We attribute the difference in this ratio to the difference in seeing conditions and fibre diameters and interpret it to mean that the young stellar population is more centrally concentrated than the old stellar population. In the dim state, the quasar continuum is detected above the 1σ level of the noise in the data at wavelengths shorter than approximately 6800 Å and the broad Hβ line is not detected at all. We can place a limit on the dim-state broad Hβ line by asking what broad Hβ fluxes could be hidden in the noise in the optical part of the spectrum; using this criterion the broad Hβ integrated flux must be less than 13.7 × 10⁻¹⁷ erg s⁻¹ cm⁻² at the 1σ level.

Table 1.

Properties of strong emission lines.

		Bright state				Dim state
		λ	FWHM	Integrated		λ	FWHM	Integrated
Line a		(Å)	(km s−1)	Flux b		(Å)	(km s−1)	Flux b
[Oii] n		– c	–	–		3729.0±0.2	499±7 d	20±1
[Neiii] n		–	–	–		3970.0±1.0	1310±60 d	10±1
H δ b		4110.0±1.0	4170±60	99±6		–	–	–
H γ b		4346.0±0.7	2800±40	290±3		–	–	–
Hβ b1 e		4867.0±0.4	1960±60	530±40		–	–	–
b2 e		4860.0±0.9	3800±200	450±40		–	–	–
b f		4864.0±0.3	2350±50	980±10		–	–	<13.7
n		–	–	–		4861.9±0.1 d	361±7 d	11±1
[Oiii] n		–	–	–		5001.0±0.4	1310±60 d	82±3
Hei b		5895.0±1.0	3200±300	200±10		–	–	–
[Oi] n		–	–	–		6363.0±0.5	670±30	5±1
[Nii] n		–	–	–		6584.0±0.1	499±7 d	50±2
Hα b1 e		6568.0±0.1	2260±20	3430±40		6563.5±0.1	2720±70	87±3
b2 e		6580.0±1.0	10200±400	1340±30		–	–	–
b f		6571.0±0.0	2420±30	4770±30		6563.5±0.1	2720±70	87±3
n		–	–	–		6564.0±0.1 d	361±7 d	34±1
[Sii] n1		–	–	–		6717.0±0.6 d	499±7 d	23±1
n2		–	–	–		6732.0±0.6 d	499±7 d	17±1

		Bright state				Dim state
		λ	FWHM	Integrated		λ	FWHM	Integrated
Line a		(Å)	(km s−1)	Flux b		(Å)	(km s−1)	Flux b
[Oii] n		– c	–	–		3729.0±0.2	499±7 d	20±1
[Neiii] n		–	–	–		3970.0±1.0	1310±60 d	10±1
H δ b		4110.0±1.0	4170±60	99±6		–	–	–
H γ b		4346.0±0.7	2800±40	290±3		–	–	–
Hβ b1 e		4867.0±0.4	1960±60	530±40		–	–	–
b2 e		4860.0±0.9	3800±200	450±40		–	–	–
b f		4864.0±0.3	2350±50	980±10		–	–	<13.7
n		–	–	–		4861.9±0.1 d	361±7 d	11±1
[Oiii] n		–	–	–		5001.0±0.4	1310±60 d	82±3
Hei b		5895.0±1.0	3200±300	200±10		–	–	–
[Oi] n		–	–	–		6363.0±0.5	670±30	5±1
[Nii] n		–	–	–		6584.0±0.1	499±7 d	50±2
Hα b1 e		6568.0±0.1	2260±20	3430±40		6563.5±0.1	2720±70	87±3
b2 e		6580.0±1.0	10200±400	1340±30		–	–	–
b f		6571.0±0.0	2420±30	4770±30		6563.5±0.1	2720±70	87±3
n		–	–	–		6564.0±0.1 d	361±7 d	34±1
[Sii] n1		–	–	–		6717.0±0.6 d	499±7 d	23±1
n2		–	–	–		6732.0±0.6 d	499±7 d	17±1

Notes. ^a b = broad component (number identifies Gaussian components summed to produce total line profile), n = narrow component. Wavelengths for lines that are in a doublet are for the long-wavelength line.

^b Integrated fluxes are in units of 10⁻¹⁷ erg s⁻¹ cm⁻².

^c The bright-state narrow lines parameters are fixed to the dim-state values.

^d This value is constrained during the spectral decomposition. See Section2.2 for a description of the constraints.

^e One of two Gaussian components.

^f Total profile, the sum of two Gaussian components.

Table 1.

Properties of strong emission lines.

		Bright state				Dim state
		λ	FWHM	Integrated		λ	FWHM	Integrated
Line a		(Å)	(km s−1)	Flux b		(Å)	(km s−1)	Flux b
[Oii] n		– c	–	–		3729.0±0.2	499±7 d	20±1
[Neiii] n		–	–	–		3970.0±1.0	1310±60 d	10±1
H δ b		4110.0±1.0	4170±60	99±6		–	–	–
H γ b		4346.0±0.7	2800±40	290±3		–	–	–
Hβ b1 e		4867.0±0.4	1960±60	530±40		–	–	–
b2 e		4860.0±0.9	3800±200	450±40		–	–	–
b f		4864.0±0.3	2350±50	980±10		–	–	<13.7
n		–	–	–		4861.9±0.1 d	361±7 d	11±1
[Oiii] n		–	–	–		5001.0±0.4	1310±60 d	82±3
Hei b		5895.0±1.0	3200±300	200±10		–	–	–
[Oi] n		–	–	–		6363.0±0.5	670±30	5±1
[Nii] n		–	–	–		6584.0±0.1	499±7 d	50±2
Hα b1 e		6568.0±0.1	2260±20	3430±40		6563.5±0.1	2720±70	87±3
b2 e		6580.0±1.0	10200±400	1340±30		–	–	–
b f		6571.0±0.0	2420±30	4770±30		6563.5±0.1	2720±70	87±3
n		–	–	–		6564.0±0.1 d	361±7 d	34±1
[Sii] n1		–	–	–		6717.0±0.6 d	499±7 d	23±1
n2		–	–	–		6732.0±0.6 d	499±7 d	17±1

		Bright state				Dim state
		λ	FWHM	Integrated		λ	FWHM	Integrated
Line a		(Å)	(km s−1)	Flux b		(Å)	(km s−1)	Flux b
[Oii] n		– c	–	–		3729.0±0.2	499±7 d	20±1
[Neiii] n		–	–	–		3970.0±1.0	1310±60 d	10±1
H δ b		4110.0±1.0	4170±60	99±6		–	–	–
H γ b		4346.0±0.7	2800±40	290±3		–	–	–
Hβ b1 e		4867.0±0.4	1960±60	530±40		–	–	–
b2 e		4860.0±0.9	3800±200	450±40		–	–	–
b f		4864.0±0.3	2350±50	980±10		–	–	<13.7
n		–	–	–		4861.9±0.1 d	361±7 d	11±1
[Oiii] n		–	–	–		5001.0±0.4	1310±60 d	82±3
Hei b		5895.0±1.0	3200±300	200±10		–	–	–
[Oi] n		–	–	–		6363.0±0.5	670±30	5±1
[Nii] n		–	–	–		6584.0±0.1	499±7 d	50±2
Hα b1 e		6568.0±0.1	2260±20	3430±40		6563.5±0.1	2720±70	87±3
b2 e		6580.0±1.0	10200±400	1340±30		–	–	–
b f		6571.0±0.0	2420±30	4770±30		6563.5±0.1	2720±70	87±3
n		–	–	–		6564.0±0.1 d	361±7 d	34±1
[Sii] n1		–	–	–		6717.0±0.6 d	499±7 d	23±1
n2		–	–	–		6732.0±0.6 d	499±7 d	17±1

Notes. ^a b = broad component (number identifies Gaussian components summed to produce total line profile), n = narrow component. Wavelengths for lines that are in a doublet are for the long-wavelength line.

^b Integrated fluxes are in units of 10⁻¹⁷ erg s⁻¹ cm⁻².

^c The bright-state narrow lines parameters are fixed to the dim-state values.

^d This value is constrained during the spectral decomposition. See Section2.2 for a description of the constraints.

^e One of two Gaussian components.

^f Total profile, the sum of two Gaussian components.

Motivated by the apparently weak power-law continuum in the dim state, we also test our model assumptions for the dim-state fit by asking whether we can obtain a similarly good decomposition without a power-law component. We perform a second fit to the dim-state spectrum, identical to our adopted dim-state decomposition described above but with the power-law component removed. Visually, the result describes the data well. We compare the Bayesian Information Criterion (BIC; Schwartz 1978), similar to the chi-squared statistic but appropriate for a comparison of models with a different number of free parameters, for this new fit and our adopted dim-state fit and find that statistically they describe the data equally well. The emission-line components of the fit are essentially unchanged compared to our adopted fit, so the result of removing the power-law continuum is that the normalizations of the stellar continuum components increase and the age of the young stellar population decreases to ∼100 Myr. Thus, the best-fitting power-law continuum in the dim state and the properties derived from it (e.g. luminosities) should be considered upper limits.

2.3 Inferred quasar properties

Using the decomposed bright-state spectrum, we measure fundamental properties of the central black hole. In the bright state, we apply the 5100 Å bolometric correction with a nonzero intercept from table 1 of Runnoe, Brotherton & Shang (2012) to the monochromatic optical luminosity λL _λ(5100 Å) = (6.3 ± 0.1) × 10⁴³ erg s⁻¹ to calculate a bolometric luminosity of L _bol = (6.8 ± 0.1) × 10⁴⁴ erg s⁻¹. The broad Hβ FWHM of 2350 ± 50 km s⁻¹, combined with the single-epoch mass scaling relationship from Vestergaard & Peterson (2006), yields M _BH = (3.6 ± 0.2) × 10⁷ M_⊙. These estimates produce an Eddington ratio of L/L _Edd = 0.15. We can also calculate the black hole mass from Hα by adopting the scaling relationship from Greene, Peng & Ludwig (2010). This calibration differs slightly from that of Vestergaard & Peterson (2006), as discussed by Cales et al. (2013), but the difference is small compared to the intrinsic scatter. This approach yields M _BH = (4.7 ± 0.1) × 10⁷ M_⊙ and L/L _Edd = 0.12. The uncertainties quoted on these mass measurements do not include the 0.4–0.5 dex scatter associated with the single-epoch mass scaling relationships. This additional source of error translates to a factor of 2.5–3 uncertainty in mass and is more representative of the uncertainty than the formal errors. Thus, the Hα and Hβ masses are consistent with each other given the large uncertainties associated with the mass scaling relationships. Using the same relationships for the dim state, λL _λ(5100 Å) = (6.3 ± 0.1) × 10⁴² erg s⁻¹, L _bol = (8.4 ± 0.1) × 10⁴³ erg s⁻¹, M _BH = (1.8 ± 0.3) × 10⁷ M_⊙, and L/L _Edd = 0.04 (the mass measurement is for Hα as Hβ is not detected).

In principle, because the black hole mass is constant, the continuum luminosity and broad-line widths should change in concert such that the single-epoch mass measurements are consistent between the bright and dim states. LaMassa et al. (2015) found this to be the case for J0159+0033. Our estimates for the Hα masses are preserved within a factor of 3, although not within the formal 1σ uncertainties.

To constrain the excitation mechanism of the narrow-line gas, we plot the observed [Oiii]/Hβ and [Nii]/Hα narrow-line ratios in a diagnostic diagram (also known as a 'BPT' diagram after Baldwin, Phillips & Terlevich 1981). The location of the measured ratios in this diagram can reveal the relative contributions to the ionizing continuum from the AGN and star formation. The line ratios from the dim-state spectrum, where the narrow-line decompositions were performed, are shown in Fig.5. We include the classification schemes of Kewley et al. (2001) and Kauffmann et al. (2003) to separate line ratios characteristic of AGN, star-forming, and composite galaxies and show as contours the distribution of SDSS galaxies from Tremonti et al. (2004) for comparison. The conclusion for J1011+5442 is clear: a hard AGN continuum is required to produce the observed narrow lines. This does not change if we adopt the alternative line ratios of [Oi] λ6300/Hα or [Sii] λλ6716, 6731/Hα in place of [Nii]/Hα.

Figure 5.

Diagnostic emission-line ratios for J1011+5442. The narrow lines in the bright-state spectral decomposition were fixed to the dim-state values, so only one position is obtained on the BPT diagram. The BPT identification schemes from Kewley et al. (2001) and Kauffmann et al. (2003) are shown as dotted and dashed lines, respectively. For comparison, grey contours show the properties of all SDSS galaxies in Tremonti et al. (2004). The emission-line ratios in J1011+5442 are consistent with a pure AGN ionizing continuum.

3 ANALYSIS

In this section, we perform a series of tests on the spectra and LINEAR+CSS light curve in order to constrain the physical mechanism responsible for the drastic changes observed in J1011+5442. We consider three physical scenarios for the observed dimming of the quasar central engine: abrupt obscuration of the central engine, intrinsic dimming due to a drop in the accretion rate (also examined as explanations for the dimming of J0159+0033 by LaMassa et al. 2015), and decay of a flare produced by the tidal disruption of a star by the central BH (proposed in the context of J0159+0033 by Merloni et al. 2015).

3.1 Obscuration of the central engine

Following LaMassa et al. (2015), we examine whether the time required for a foreground object in a bound orbit around the BH to move in front of the continuum source and BLR can be short enough to match the observed transition time. We adopt equation 4 of LaMassa et al. (2015),

\begin{equation} t_{{\rm cross}} = 0.07 \left[ \frac{r_{{\rm orb}}}{1\,\textrm{lt} - \textrm{d}}\right]^{3/2} M_{8}^{-1/2} \textrm{arcsin}\left[ \frac{r_{{\rm src}}}{r_{{\rm orb}}}\right]\textrm{yr}, \end{equation}

(3)

which gives the time for an object at an orbital radius r _orb to traverse a region extending to radius r _src around a black hole of mass M ₈ =M/10⁸M_⊙. The most conservative case, which minimizes the crossing time, corresponds to the minimum plausible value of r _src and a value of r _orb that is determined graphically. Thus, we begin by assuming that r _src is the radius derived from the 'Clean2+ExtCorr' radius–luminosity relation (R _BLR; Bentz et al. 2013), i.e. an emissivity-weighted radius that encompasses the bulk of the broad-line emission. For J1011+5442, R _BLR = 28 light-days; assuming that r _src =R _BLR, we estimate a minimum crossing time of t _cross ∼ 21 yr with r _orb = 34 light days. This minimizes the crossing time since we have effectively assumed that the obscuring object is orbiting within the BLR and is large enough to obscure its brightest parts. The lower limit on the minimum crossing time is t _cross ∼ 9 yr, calculated by subtracting the 0.13 dex scatter in the R–L relationship from log(R _BLR), increasing the mass by a factor of 2.5 to account for the uncertainty in the mass scaling relationship, and recalculating the crossing time under the same assumptions about r _src and r _orb. However, more realistically, the obscuring object should orbit outside the BLR, whose outer edge has a radius that is several times larger than emissivity-weighted radius R _BLR. Reverberation mapping of infrared emission from illuminated dust indicates that the BLR is contained within ∼3R _BLR (e.g. Baldwin et al. 2004; Suganuma et al. 2006; Barth et al. 2013, see LaMassa et al. 2015 for a more detailed justification of this assumption), so we set r _orb =r _src = 3R _BLR. This leads to a crossing time of t _cross ∼ 31 yr. To obscure all of the BLR, or for screens located farther from the source, crossing times can easily reach hundreds of years. In all cases, these time-scales are far too long to explain the dimming seen here.

If the power-law continuum is detected, we can carry out a complementary test by comparing the ratio of fluxes from the dim- and bright-state spectra at specific wavelengths and asking whether they are all consistent with the same amount of reddening. The ratio of the Hα flux between the dim and bright states is 0.018 ± 0.001, which requires E(B −V) = 1.744 in order to be explained by extinction. The flux ratio of Hα/Hβ changes by >1.3 from the bright to dim state, implying E(B −V) > 0.248, which is consistent with the decline in the Hα flux. The ratio of the quasar continua from the two epochs at 6563 Å is 0.15 ± 0.08. which is not consistent with the amount of dimming measured at the same wavelengths in the Hα line. The Hβ line and nearby continuum should also demonstrate the same amount of attenuation, but this line ratio is <0.014 whereas the 5100 Å continuum is 0.100 ± 0.003. Finally, at 3500 Å the amount of reddening needed to attenuate Hα produces a continuum flux that is about a factor of 100 too weak. Thus, if the continuum is detected from the quasar, the relative dimming of the emission lines and the continuum is not consistent with reddening. If, on the other hand, the power-law continuum is an upper limit, this discrepancy is alleviated and the real problem for the reddening scenario is the time-scale for obscuration to set in.

Additional evidence against the reddening scenario can be found in the broad Hα line width; the line profile changes such that Hα is broader in the dim state, suggesting that it is emitted nearer to the central engine which would presumably suffer from the most extinction. Based on all these calculations, we disfavour reddening by a foreground screen as an explanation of the changes in this object.

3.2 Intrinsic dimming of the continuum source

An alternative explanation for the observed variability is that a rapid change in accretion rate is responsible for the dimming in the continuum and broad emission lines. We evaluate whether this might be possible by calculating the accretion time-scale for the inner part of the accretion disc where most of the ionizing radiation is generated (R < 10R _g , where R _g  =GM _BH/c ² is the gravitational radius). We follow equation 5 of LaMassa et al. (2015) for the inflow time-scale (i.e. the time for a parcel of gas to move radially from a given radius in the accretion disc to the centre, also known as the viscous or radial drift time-scale), adopting their fiducial value for the viscosity parameter (Shakura & Sunyaev 1973), but substituting our measured values of M ₈ = 0.36 and λ_Edd = 0.15. This gives an inflow time of t _infl = 0.25 yr, or equivalently three months. This time-scale is short enough that a drop in accretion can easily explain the changes in the continuum that are observed in J1011+5442. The inflow time-scale we estimate here is considerably shorter than what LaMassa et al. (2015) estimated for J0159+0033 because the two quasars have considerably different masses and accretion rates (in the bright state, J1011+5442 has M ₈ = 0.36 and λ_Edd = 0.15 versus M ₈ = 1.7 and λ_Edd = 0.04 for J1059+0033).

It is interesting to note that the continuum throughout the optical band dimmed within approximately 500 d, as well as the far-UV continuum at ∼1 Ry, which is responsible for the ionization of hydrogen and the production of the Balmer lines. As noted by LaMassa et al. (2015), these changes may be driven by a drop in the luminosity of the inner accretion disc even though the optical continuum is produced in a region of the accretion disc that is much larger than 10R _g . This is because the optical continuum may result from illumination from the inner disc, as suggested by the results of reverberation studies (e.g. Krolik et al. 1991; Cackett, Horne & Winkler 2007). Similarly, since the 'radius of the broad-line region', R _BLR, is only 28 light days, the luminosity of the broad Balmer emission lines will track the decline of the Lyman continuum with a lag that is less than a month. Therefore, the transformation of J1011+5442 can be accomplished by a rapid drop in the far-UV continuum luminosity resulting from an abrupt decline in the accretion rate. The mechanism behind this change in the accretion rate, however, is unknown. Tantalizing as this scenario may be, it must remain tentative until a plausible mechanism for abrupt changes in accretion rate is identified.

3.3 Tidal disruption events

A decaying flare from a TDE was proposed as a possible mechanism for the changes observed in J0159+0033 by Merloni et al. (2015). Specifically, J1059+0033 was initially observed in a lowly accreting state, but brightened dramatically before spectroscopy revealed the presence of broad lines, which later disappeared. We use the LINEAR+CSS light curve to evaluate if the TDE scenario can explain J1011+5442. In this case, there is no dramatic brightening of the light curve, so the entire event is not covered. Rather, the hypothetical TDE is already underway at the start of our observations, so we resolve only part of the peak of the light curve followed by its decline.

For a more quantitative investigation, our first step is to derive the light curve of the variable component by subtracting the flux of the host galaxy. Although the TDSS spectrum provides a good measure of the host emission, the 2.0 arcsec aperture likely misses some of the flux. Therefore, instead of using the TDSS spectrum, we assume that the faintest point in the CSS light curve provides an upper limit on the host flux, which we subtract from all points in the light curve. This step represents the largest uncertainty in characterizing the decay of the light curve and we discuss the implications of this choice in more detail below.

Analytic calculations (Phinney 1989) and numerical modelling (e.g. Evans & Kochanek 1989; Ayal, Livio & Piran 2000) of TDEs predict that the light curve of a TDE will decline as t ^α, where α = −5/3. In principle, this t ^−5/3 decay law that is traditionally associated with TDEs actually describes the rate at which material will return to the vicinity of the black hole, which is likely not equal to the accretion rate. In fact, Lodato & Rossi (2011) predict different power-law indices for light curves in different electromagnetic bands. Nonetheless, the majority of the TDEs claimed to date do exhibit this decay law in their light curves (e.g. Gezari et al. 2009). To test whether the declining light curve of J1011+5442 is consistent with this expectation, we write the evolution of its V-band magnitude as

\begin{equation} V(t) = V_{0} + \alpha \,\textrm{log}\left(\frac{t-t_{0}}{1\,\textrm{yr}}\right), \end{equation}

(4)

where t ₀ indicates the beginning of the decline in the light curve, observationally constrained to be between 2009.4 and 2009.9. To fit the light curve, we adopt t ₀ = 2009.65 and use the range in t ₀ to determine the uncertainty on α. We find |$\alpha = -1.3^{+0.8}_{-0.7}$|⁠, which is formally consistent with −5/3; both curves are presented with the host-subtracted light curve in Fig.6. We note that, for some specific values of t ₀ within 2009.4–2009.9, it is possible to obtain slopes that are inconsistent with −5/3. We investigate the impact of the host subtraction by varying the host magnitude and repeating the fit to the light curve. Adopting a host magnitude of V = 19.26, determined from the host spectral components from the SDSS spectrum (which has the largest fibre and should capture more of the host light than the TDSS observation), we obtain |$\alpha = -1.0^{+0.6}_{-0.5}$|⁠, which is not consistent with the −5/3 slope. Thus, the results of this test suggest that the decay of the light curve may be consistent with a slope of −5/3, but other values are possible as well.

Figure 6.

The Catalina Sky Survey light curve of the variable emission with the host subtracted. Red points indicate data that were included in fitting the TDE model, while other points are shown with grey error bars only. The grey region indicates the observational constraints on t ₀ and the dashed grey line is the adopted value. The solid black line represents the best-fitting theoretically based TDE model, where |$\alpha = -1.3^{+0.8}_{-0.7}$|⁠, and the dotted line shows the same model with the slope fixed at α = −5/3. The dashed line shows the empirically based best-fitting exponential model for the light curve from equation (5), where β = 0.4 ± 0.1.

Empirically, the declining luminosity of TDE candidates is sometimes better described as an exponential decay. This was pointed out by Holoien et al. (2014, 2015), who studied events whose broad Balmer and Heii emission lines and spectral energy distribution evolved in a manner inconsistent with supernovae. We therefore also evaluate whether the light curve of J1011+5442 is consistent with the light curves of observed TDE candidates by describing the V-band magnitude as

\begin{equation} V(t) = V_{0} + \textrm{log}\left[ \textrm{exp}\left(-\frac{t-t_{0}}{\beta }\right)\right], \end{equation}

(5)

where t ₀ still indicates the beginning of the decay, β is the e-folding time (treated as a free parameter), and the normalization V ₀ may be different from the value determined when fitting equation (4). The exponential fit is not very sensitive to the value of t ₀, so we adopt t ₀ = 2009.65 with no variation. We find best-fitting parameters of V ₀ = −18.5 ± 0.2 mag and β = 0.4 ± 0.1 yr. This is not consistent with the best-fitting values of β ≃ 0.1 − 0.2 yr that describe the observed TDE candidates (Holoien et al. 2014, 2015). Statistically, we find that the exponential fit is not preferred to the traditional t ^α, where Δ(BIC) < 1 for any pair of models that we consider. This is likely the result of the fact that the decaying phase of the light curve comprises a relatively small number of data points with substantial error bars. Thus, no model stands out for providing a better fit to the data compared to the other models.

We can also calculate the amount of matter accreted by the black hole during the bright state and compare it to reasonable values for stars that might be tidally disrupted. That is to say, the amount of matter accreted during the high state cannot exceed the mass of the disrupted star. Assuming an efficiency of η = 0.1, and adopting the measured bolometric luminosity from the high state, the black hole accreted 0.7 M_⊙ in the ∼6 yr that it was in the bright state.

Although the above tests appear to support the TDE interpretation, the overall shape of the light curve does not. The most recent simulations for feeding black holes with tidal disruption of stars show that the light curve of a TDE has a very narrow peak Guillochon & Ramirez-Ruiz (2013). From their Fig.5, a time-scale of a few months is a generous upper limit on the bright peak in the light curve of a TDE. Similarly, the observed light curves of strong TDE candidates (e.g. Gezari et al. 2009) remain in the bright state only for a very short time, time-scales of the order of months. In contrast, J1011+5442 is in the bright state for at least ∼6 yr in the rest frame.

The bright-state emission lines are difficult to explain in the TDE scenario. The observed broad emission lines suggest a BLR that is too extended and massive to have been deposited by the recent tidal disruption of a single star. The narrow lines, which do not change, also present challenges, requiring a hard ionizing continuum and gas at even larger radii. Merloni et al. (2015) discuss these points and relevant calculations in detail, but the issue remains that, while a TDE can illuminate an existing BLR, it cannot distribute the necessary gas around the black hole or produce narrow lines coincident with the peak of the light curve. Based on all these lines of evidence, we therefore reject the TDE scenario to describe J1011+5442.

4 SUMMARY AND DISCUSSION

In this work, we report the discovery and analysis of the z = 0.246 changing-look quasar, SDSS J101152.98+544206.4. In the bright state, observed in early 2003 as part of the SDSS, J1011+5442 is listed in the SDSS quasar catalogues up to DR7 and the optical spectrum of J1011+5442 exhibits the broad emission lines and a blue continuum of a typical broad-lined quasar. From this spectrum, we determine L _bol = (6.8 ± 0.1) × 10⁴⁴ erg s⁻¹, M _BH = (3.6 ± 0.2) × 10⁷ M_⊙, and λ_Edd = 0.15. The TDSS targeted the object for observations in early 2015 and found it in a dim state: the quasar continuum was undetected at the 1σ level at wavelengths longer than ∼6800 Å, and of the broad emission lines, only Hα was still present, but at a much weaker level than was observed during the bright state in 2003. The 2006–2013 CSS light curve reveals that the transition likely occurred over ∼500 d in the rest frame around 2010.

We test several physical scenarios to explain the dramatic changes in J1011+5442's spectrum. In some cases (typically in Seyfert 1.8–1.9 galaxies as we describe below), variable obscuration of the central engine can explain changing-look AGN, but this mechanism is disfavoured in this case. The time-scale for an intervening cloud of dust to obscure the BLR is too long, of the order of 30 yr, compared to the measured time-scale of ∼500 d. An additional test of the obscuration scenario would be to ask whether the observed change in the quasar continuum is consistent with the amount of reddening needed to produce the change in Hα. We attempt this test, but the results are inconclusive because the quasar continuum is not robustly detected in the optical. Hubble Space Telescope observations in the ultraviolet, where the power-law continuum rises, would be required to perform this test conclusively. The observations of J1011+5442 cannot be explained by a TDE. Although the decay rate of the light curve is consistent with theoretical expectations for TDEs and the 0.7 M_⊙ accreted while J1011+5442 was in the bright state could have been deposited by disruption of a star, the bright state has a minimum duration that is far longer than any putative observed TDE or theoretical TDE model. An abrupt change in the accretion rate can drive the changing look of J1011+5442. The time-scale for material to be transported through the inner accretion disc and for the disc to respond to changes in the accretion rate is of the order of months, consistent with the observed change, making this a viable process.

In the context of recently discovered changing-look quasars (LaMassa et al. 2015; Ruan et al. 2015), J1011+5442 has similar luminosity, redshift, and transition time-scale. All recently discovered changing-look quasars cluster in the upper right corner of the luminosity–redshift diagram of LaMassa et al. (2015, see their fig. 1). J1011+5442 does, however, have a notably smaller black hole mass and larger Eddington ratio compared to J0159+0033, which lead to a much shorter inflow time-scale in the inner accretion disc (months compared to decades for J0159+0033). As a result, a large, abrupt fluctuation in the accretion rate is a viable explanation of the variability of J1011+5442, whereas with J0159+0033 there is some potential for ambiguity. Elitzur, Ho & Trump (2014) describe an accretion-rate-driven paradigm for the evolution between spectral types. This is based on the sample of intermediate-type AGNs from Stern & Laor (2012), where a decreasing fraction of the bolometric luminosity is converted to broad-line emission as the accretion rate, characterized by |$L_{{\rm bol}}/M_{{\rm BH}}^{2/3}$|⁠, drops. J1059+0033 is consistent with the expected values of |$L_{{\rm bol}}/M_{{\rm BH}}^{2/3}$| for this sequence, but it is not possible to ascribe the spectral changes to changes in the accretion rate with complete certainty because the inflow time-scale is so long, whereas J1011+5442 is a fairly unambiguous case because of the shorter inflow time-scale. Thus, the changing-look quasars appear to be more luminous analogues of some of the Seyfert 1.8–1.9 galaxies studied by Goodrich (1989, 1990, 1995) and later Trippe et al. (2010). As Goodrich and Trippe et al. conclude from a variety of observational tests, the variability of the continuum and broad Balmer lines of some objects could be attributed to intrinsic variations while in others it was caused by changes in the reddening.

Our understanding of changing-look quasars will benefit substantially from continued optical spectroscopic monitoring as well as X-ray observations. X-ray spectroscopy was important in rejecting the obscuration scenario for J0159+0033 and can serve the same purpose for the other changing-look quasars. Spectroscopic monitoring can serve at least two purposes. First, it can reveal whether the quasar central engine turns on again because X-ray emission originates near the black hole and acts as an instantaneous tracer of accretion. As we discover and monitor larger numbers of such objects, we will be able to constrain the duration of the 'off-state'.' Moreover, if the quasar returns, this would be a strong indication that the previous 'on-state' was not caused by a TDE since the rate of such events is expected to be ∼10⁻⁵–10⁻⁴ galaxy⁻¹ yr⁻¹ (both on theoretical and observational grounds, e.g. Donley et al. 2002; Merritt & Poon 2004; Gezari et al. 2008; Maksym, Ulmer & Eracleous 2010; van Velzen & Farrar 2014; Holoien et al. 2015). Secondly, as LaMassa et al. (2015) remarked, we may observe a decline in the narrow emission lines, starting potentially as early as several years after the decline of the continuum. The precise onset of the response will depend on the structure of the NLR in J1011+5442, but can happen more quickly than imagined given that the traditional kiloparsec-scale radius of the NLR describes the outer extent rather than the innermost radius of the gas (see discussions on narrow-line variability in e.g. Denney et al. 2014; LaMassa et al. 2015). This would afford an additional test of the intrinsic dimming hypothesis and it would allow us to probe the structure of the NLR. There are, in fact, specific predictions of how the narrow lines would respond to a sharp drop in the ionizing continuum, based on time-dependent photoionization calculations by Binette & Robinson (1987) and Eracleous, Livio & Binette (1995). Among the noteworthy predictions are the rapid response of the optical [Oiii] doublet (the line flux tracks the decline of the ionizing continuum with a lag of only a few years) and the very slow response of the Balmer lines (they remain fairly steady for decades to centuries after the ionizing continuum has faded).

The TDSS offers a promising way of discovering substantial numbers of changing-look quasars because it will revisit several thousand objects with previous spectra from the SDSS. The two changing-look quasars discovered within the first year of the TDSS may be the tip of the iceberg. By design, however, the TDSS is unlikely to find transitions from the 'off' to the 'on' state, which are just as interesting and valuable in their own right, because the TDSS targets are known AGN.

We thank the referee, Benjamin Shappee, for helpful comments and suggestions. ARL acknowledges partial financial support from the DIULS Regular project PR15143.

Funding for the Sloan Digital Sky Survey IV has been provided by the Alfred P. Sloan Foundation and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High-Performance Computing at the University of Utah. The SDSS web site is www.sdss.org.

SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, Carnegie Institution for Science, Carnegie Mellon University, the Chilean Participation Group, Harvard-Smithsonian Center for Astrophysics, Instituto de Astrofísica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU) / University of Tokyo, Lawrence Berkeley National Laboratory, Leibniz Institut fr Astrophysik Potsdam (AIP), Max-Planck-Institut fr Astrophysik (MPA, Garching), Max-Planck-Institut fr Extraterrestrische Physik (MPE), Max-Planck-Institut fr Astronomie (MPIA, Heidelberg), National Astronomical Observatory of China, New Mexico State University, New York University, University of Notre Dame, Observatrio Nacional do Brasil, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group, Universidad Nacional Autnoma de México, University of Arizona, University of Colorado Boulder, University of Portsmouth, University of Utah, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University.

The CSS survey is funded by the National Aeronautics and Space Administration under Grant No. NNG05GF22G issued through the Science Mission Directorate Near-Earth Objects Observations Program. The CRTS survey is supported by the US National Science Foundation under grants AST-0909182.

The LINEAR program is funded by the National Aeronautics and Space Administration at MIT Lincoln Laboratory under Air Force Contract FA8721- 05-C-0002. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the United States Government.

This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration.

REFERENCES

et al.

2009

ApJ

698

1367

et al.

2014

MNRAS

445

3263

et al.

2003

MNRAS

346

1055

1994

ASP Conf. Ser. Vol. 61, Astronomical Data Analysis Software and Systems III

Astron. Soc. Pac.

San Francisco

437

et al.

2013a

MNRAS

432

1351

et al.

2013b

MNRAS

436

3581

1984

Sov. Astron. Lett.

10

335

et al.

2015

MNRAS

452

69

1989

Proc. IAU Symp. 136, The Center of the Galaxy

Kluwer

Dordrecht

543

© 2015 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society

© 2015 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society

robertsdoccom.blogspot.com

Source: https://academic.oup.com/mnras/article/455/2/1691/1113198

Share this post