Two New "Turn-off" Changing-look Active Galactic Nuclei and Implication on "Partially Obscured" AGNs

As a challenge to the widely accepted picture of active galactic nuclei (AGNs; e.g., Antonucci 1993; Shen & Ho 2014), the so-called "changing-look" (CL) phenomenon is a hot topic in modern astronomy. Based on the generally accepted unified model (see Antonucci 1993 for a review), the observed Type-I and -II spectra are explained by the orientation effect caused by the dust torus. Type-I AGNs with both broad ($\mathrm{FWHM}\gt 1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$) and narrow ($\mathrm{FWHM}\sim {10}^{2}\,\mathrm{km}\,{{\rm{s}}}^{-1}$) Balmer emission lines are believed to be observed face-on, while Type-II AGNs without the broad Balmer emission lines are observed edge-on. In the CL phenomenon, an AGN changes its optical spectral type on a timescale of the order of years, although Trakhtenbrot et al. (2019) recently claimed an identification of a CL phenomenon on a timescale of months in 1ES 1927+654 by their high-cadence spectroscopic monitoring.

To date, there are only ∼70 changing-look active galactic nuclei (CL-AGNs) discovered by repeat and sparse spectroscopic observations, although both turn-on and turn-off transitions have been identified (e.g., Shapovalova et al. 2010; Shappee et al. 2014b; LaMassa et al. 2015; MacLeod et al. 2016; McElroy et al. 2016; Ruan et al. 2016; Runnoe et al. 2016; Gezari et al. 2017; Stern et al. 2018; Wang et al. 2018; Yang et al. 2018; MacLeod et al. 2019). By reporting six turn-on CL-AGNs discovered during the first nine months operation of the Zwicky Transient Facility survey, Frederick et al. (2019) recently proposed a new class of changing-look low-ionization nuclear emission-line regions (CL-LINERs), whose spectra in the quiescent state can be classified as a LINER with weak emission lines. An additional case of CL-LINER with a rapid "turn-on" transition to type-1 AGN was recently discovered in the Sloan Digital Sky Survey (SDSS) J111536.57+054449.7 by Yan et al. (2019).

In contrast to the CL phenomenon discovered through X-rays (e.g., Risaliti et al. 2009), which is generally believed to be caused by a large change of the line-of-sight absorption column (e.g., Matt et al. 2003; Piconcelli et al. 2007; Ricci et al. 2016), there is accumulating evidence supporting that the optical CL phenomenon is likely due to a variation in the accretion rate of a supermassive black hole (SMBH). Such variation results from either a viscous radial inflow or from disk instability (e.g., Gezari et al. 2017; Sheng et al. 2017; Wang et al. 2018; Yang et al. 2018), although an explanation of accelerating outflow launched from the central SMBH cannot be entirely excluded (e.g., Shapovalova et al. 2010). In the instability scenario, a thermal instability with a shorter timescale in the optical region is additionally involved to solve the timescale problem (e.g., Husemann et al. 2016; Lawrence 2018). MacLeod et al. (2019) recently proposed that the optical CL phenomenon can be explained by an accretion supported broad-line region (BLR) in the disk–wind model (e.g., Nicastro 2000; Elitzur & Ho 2009; Elitzur et al. 2014), in which the BLR appears or disappears when the luminosity is above or below the critical one.

In this paper, we report an identification of two new CL-AGNs with a turn-off transition by repeat spectroscopy of the CL-AGN candidates recently selected by MacLeod et al. (2019). The paper is organized as follows. Section 2 describes the sample used. The observations and spectral analysis are presented in Sections 3 and 4, respectively. A conclusion and discussion focusing on the dim state of a sample of CL-AGNs are presented in the last section. A ΛCDM cosmology with parameters ${H}_{0}=70\,\mathrm{km}\,{{\rm{s}}}^{-1}\ {\mathrm{Mpc}}^{-1}$, Ω_m = 0.3, and Ω_Λ = 0.7 is adopted throughout the paper.

MacLeod et al. (2019) recently identified 17 new CL quasars (CLQs) through a follow-up spectroscopy to the highly variable SDSS DR7 spectroscopically identified quasars. The variability is required to be $| {\rm{\Delta }}g| \gt 1$ mag and $| {\rm{\Delta }}r| \gt 0.5$ mag by comparing the photometric measurements between the SDSS DR10 and Pan-STARRS (PS1; Kaiser et al. 2002). With the spectroscopic identifications, the authors claimed a CLQ confirmation rate of ≥20%. A catalog of more than 200 highly variable quasars was additionally released in MacLeod et al. (2019) for future follow-up spectroscopic identifications of new CL-AGNs. To ensure the Hβ emission line in the observer frame is within the optical wavelength region, the redshifts of the candidates are limited to those smaller than 0.83.

We performed a follow-up spectroscopic observation program by the 2.16 m telescope at Xinglong Observatory on a subsample of the highly variable quasars catalog given by MacLeod et al. (2019). After taking into account both the celestial location and the brightness of the candidates, there is a total of only five candidates available for the telescope before July.

The follow-up spectroscopic observations and data reductions of the five CLQ candidates are described in this section.

3.1. Observations

3.2. Data Reduction

3.3. Identification of Changing-look Phenomenon

With our follow-up spectroscopy, a CL phenomenon with a turn-off transition can be clearly identified in two quasars: SDSS J104705.16+544405.8 and SDSS J120447.91+170256.8 (hereafter SDSS J1047+5444 and SDSS J1204+1702 for short). Figures 1 and 2 compare the SDSS and Xinglong spectra for the two CLQs and the three non-CLQs, respectively. The S/N of the Xinglong spectrum of SDSS J094443.08+580953.2 is too low to give any meaningful result on this object. In the comparison, the spectra taken by SDSS are convolved with a Gaussian profile to match the spectral resolution of the Xinglong spectra. The flux level of each Xinglong spectrum is scaled by a factor determined by requiring the modeled total [O iii]λ5007 line flux equals to that of the corresponding SDSS spectrum (see Section 4).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

One can see clearly from the comparison a turn-off type transition from classical type-1 to type-1.9 in both quasars, i.e., SDSS J1047+5444 and SDSS J1204+1702. The Hβ and Hγ broad components almost disappear in both Xinglong spectra, along with a significant weakening of both Hα broad emission and AGN's featureless continuum. In fact, by a direct integration over a proper wavelength range on the residual spectra, the relative variation of the Hβ broad emission line ${\rm{\Delta }}f/{f}_{\mathrm{SDSS}}$ is estimated to to be ∼−0.72 and −0.67 for SDSS J1047+5444 and SDSS J1204+1702, respectively, where f_SDSS is the line flux obtained from the SDSS spectra and ${\rm{\Delta }}f={f}_{\mathrm{Xionglong}}\mbox{--}{f}_{\mathrm{SDSS}}$. The value of ${\rm{\Delta }}f/{f}_{\mathrm{SDSS}}$ is, however, as low as −0.01, 0.14, and 0.02 for the other three non-CLQs. In SDSS J1047+5444, We argue that the continuum of the Xinglong spectrum is changed to be dominated by the host stellar emission, with a 4000 Å break due to the stellar metal absorptions and a marginally detected Mg ib (5176 Å) absorption feature marked in Figure 1.

Here, we perform a spectral analysis following Wang et al. (2018 and references therein) to help explain the turn-off type transitions occurring in the two CLQs.

4.1. AGN Continuum and Stellar Feature Removal

First, we model the continuum of each of the four spectra by a linear combination of the following components: (1) an AGN's power-law continuum; (2) a template of both high-order Balmer emission lines and a Balmer continuum from the BLR; (3) a template of both optical and ultraviolet Fe ii complex; (4) a host-galaxy template with an age of 5 Gyr extracted from the single stellar population (SSP) spectral library given in Bruzual & Charlot (2003); and (5) an intrinsic extinction due to the host galaxy described by a galactic extinction curve with R_V = 3.1. We use the empirical optical Fe ii template provided by Veron-Cetty et al. (2004) and the theoretical template by Bruhweiler & Verner (2008) to model the optical and ultraviolet Fe ii complex, respectively. The line width of the template is fixed in advance to be that of the broad component of Hβ, which is determined by our line profile modeling (see below).

The emission from a partially optically thick cloud with an electron temperature of T_e = 1.0 × 10⁴ K is adopted to model the Balmer continuum ${f}_{\lambda }^{\mathrm{BC}}$ by following Dietrich et al. (2002); see also Malkan & Sargent (1982):

$$\begin{eqnarray}&&{f}_{\lambda }^{\mathrm{BC}}={f}_{\lambda }^{\mathrm{BE}}{B}_{\lambda }({T}_{e})(1-{e}^{-\tau }),\lambda \leqslant {\lambda }_{\mathrm{BE}},\end{eqnarray} \tag{ 1 }$$

where ${f}_{\lambda }^{\mathrm{BE}}$ is the continuum flux at the Balmer edge λ_BE = 3646 Å and ${B}_{\lambda }({T}_{e})$ is the Planck function. τ_λ is the optical depth at wavelength λ, which is related to the one at the Balmer edge τ_BE as ${\tau }_{\lambda }={\tau }_{\mathrm{BE}}{(\lambda /{\lambda }_{\mathrm{BE}})}^{3}$. A typical value of τ_BE = 0.5 is adopted in our modeling of the continuum.

We model the high-order Balmer lines (i.e., H7–H50) by the case B recombination model with an electron temperature of ${T}_{e}=1.5\times {10}^{4}\,{\rm{K}}$ and an electron density of ${n}_{e}={10}^{8\mbox{--}10}\ {\mathrm{cm}}^{-3}$ (Storey & Hummer 1995). The widths of these high-order Balmer lines are, again, determined in advance according to the line profile modeling of the Hβ broad emission (see below).

A χ² minimization is performed iteratively over the whole spectroscopic wavelength range, except for the regions with known emission lines (e.g., Hα, Hβ, Hγ, Hδ, [S ii]λλ6716,6731, [N ii]λλ6548,6583, [O i]λ6300, [O iii]λλ4959,5007, [O ii]λ3727, [Ne iii]λ3869, and [Ne v]λ3426). For both SDSS spectra being typical of a type-I AGN, the underlying stellar emission is failed to be modeled because the continuum is entirely dominated by the AGN's featureless emission. The modeling of the underlying stellar emission is also failed in the Xinglong spectrum of SDSS J1204+1702 due to the poor S/N of its continuum. In contrast, the Xinglong spectrum of SDSS J1047+5444 shows a continuum being typical of an intermediate-type AGN with weak or negligible emission from the central engine, although our modeling based on a 5 Gyr old SSP returns a poor reproduction of the continuum. In addition, our modeling suggests that the optical Fe ii complex is too weak to be modeled in all the four spectra. The removal of the modeled continuum is illustrated in Figure 3 for the four spectra.

Zoom In Zoom Out Reset image size

4.2. Line Profile Modeling

For each emission-line-isolated spectrum, the emission-line profiles are modeled on both Hα and Hβ regions by the SPECFIT task (Kriss 1994) in the IRAF package.

In the profile modeling, each emission line is reproduced by a set of Gaussian profiles. By taking the poor continuum removal described above into account, a local linear continuum is additionally used for the Xinglong spectrum of SDSS J1047+5444 to reproduce the line profiles in the Hβ region. The line flux ratios of the [O iii]λλ4959, 5007 and [N ii]λλ6548, 6583 doublets are fixed to their theoretical values, i.e., 1:3. The line widths of both Hα and Hβ broad components are measured directly on the residual profiles by the SPLOT task in the IRAF package after subtracting the modeled narrow components from the observed profiles. The line modelings are shown in the left and right panels of Figure 4 for the Hβ and Hα regions, respectively. The results of the profile modeling are listed in Table 2. No intrinsic extinction correction is applied to all the derived line fluxes, because the Balmer decrements of the narrow components determined from the SDSS spectra are as small as ${\rm{H}}\alpha /{\rm{H}}\beta =2.87\pm 0.17$ and 1.43 ± 0.14 for SDSS J1047+5444 and SDSS J1204+1702, respectively. All the errors reported in the table correspond to the 1σ significance level after taking into account the proper error propagation.

Zoom In Zoom Out Reset image size

Table 2.  Spectral Measurements and Analysis

SDSS ID	Epoch	AGN type	$F([{\rm{O}}\,{\rm{III}}]\lambda 5007)$	$F({\rm{H}}{\beta }_{{\rm{b}}})$	FWHM(H${\beta }_{{\rm{b}}}$)	F(Hαb)	FWHM(Hαb)	${M}_{\mathrm{BH}}/{M}_{\odot }$	${L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$
			$({10}^{-15}\ \mathrm{erg}\ {{\rm{s}}}^{-1}\ {\mathrm{cm}}^{-2})$		$(\mathrm{km}\,{{\rm{s}}}^{-1})$	$({10}^{-15}\ \mathrm{erg}\ {{\rm{s}}}^{-1}\ {\mathrm{cm}}^{-2})$	$(\mathrm{km}\,{{\rm{s}}}^{-1})$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
J1047+5444	2002/04/04	1.0	13.5 ± 0.02	24.5 ± 1.1	2600 ± 120	102.5 ± 1.1	1970 ± 80	4.0 × 107	0.41
	2019/04/14	1.9	9.1 ± 0.2	⋯	⋯	18.7 ± 0.4	2940 ± 60	3.6 × 107	0.15
J1204+1702	2007/04/17	1.0	15.2 ± 0.3	40.0 ± 1.8	6720 ± 200	115.5 ± 1.0	4920 ± 50	3.7 × 108	0.09
	2019/04/14	1.8/1.9	14.7 ± 0.9	7.1 ± 0.4	6810 ± 600	35.7 ± 1.0	3140 ± 130	1.0 × 108	0.12

Download table as:  ASCIITypeset image

4.3. Black Hole Mass and Eddington Ratio

Based on the line profile modelings, the SMBH viral mass (M_BH) and Eddington ratio ${L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$ (where ${L}_{\mathrm{Edd}}=1.26\,\times {10}^{38}{M}_{\mathrm{BH}}/{M}_{\odot }\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ is the Eddington luminosity) are estimated for the two CLQs from the single-epoch spectroscopy through several well-established calibrated relationships (e.g., Kaspi et al. 2000, 2005; Wu et al. 2004; Peterson & Bentz 2006; Marziani & Sulentic 2012; Du et al. 2014, 2015; Peterson 2014; Wang et al. 2014).

Our estimations of both M_BH and ${L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$ are based on the modeled broad Hα emission lines. Greene & Ho (2005) provided a calibration of

$$\begin{eqnarray}&&{M}_{\mathrm{BH}}=3.0\times {10}^{6}{\left(\displaystyle \frac{{L}_{{\rm{H}}\alpha }}{{10}^{42}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{0.45}{\left(\displaystyle \frac{\mathrm{FWHM}({\rm{H}}\alpha )}{1000\mathrm{km}{{\rm{s}}}^{-1}}\right)}^{2.06}\,{M}_{\odot }\end{eqnarray} \tag{ 2 }$$

to estimate M_BH. To obtain an estimation of ${L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$, we derive the bolometric luminosity L_bol from the standard bolometric correction ${L}_{\mathrm{bol}}=9\lambda {L}_{\lambda }(5100\,\mathring{\rm A} )$ (e.g., Kaspi et al. 2000), where L_λ(5100 Å) is the AGN's specific continuum luminosity at 5100 Å, which can be inferred from the Hα broad-line luminosity through the calibration (Greene & Ho 2005)

$$\begin{eqnarray}&&\lambda {L}_{\lambda }(5100\,\mathring{\rm A} )=2.4\times {10}^{43}{\left(\displaystyle \frac{{L}_{{\rm{H}}\alpha }}{{10}^{42}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{0.86}.\end{eqnarray} \tag{ 3 }$$

The estimated ${M}_{\mathrm{BH}}$ and ${L}_{\mathrm{bol}}/{L}_{\mathrm{Edd}}$ are tabulated in Columns (9) and (10) in Table 2, respectively. For each object, the used Hα line flux measured from the Xinglong spectrum is scaled by a factor determined by equaling the total [O iii]$\lambda 5007$ line flux to that of the corresponding SDSS spectroscopy.

For SDSS J1047+5444, the two spectra taken at different epochs return consistent estimations of the M_BH. The corresponding L/L_Edd decreases from 0.41 to 0.15. For SDSS J1204+1702, however, we obtain a roughly constant L/L_Edd when the object is at the "on" and "off" states. We argue that the invariable L/L_Edd is due to the difference in the estimated M_BH, in which the M_BH determined from the SDSS spectrum is about four times larger than that from the Xinglong spectrum. By adopting the M_BH from the SDSS spectrum, similar as in SDSS J1047+5444, the corresponding L/L_Edd in fact decreases from 0.09 to 0.02.

By performing a follow-up spectroscopy on five CL-AGN candidates recently selected by MacLeod et al. (2019), we identify two new CL quasars, i.e., SDSS J1047+5444 and SDSS J1204+1702, both with a "turn-off" type transition, when the new spectra taken by the 2.16 m telescope in Xinglong Observatory are compared to the SDSS archival spectra.

With the increasing number of the identified CL-AGNs, there is accumulating evidence supporting that the change of SMBH's accretion rate is the physical origin of the CL phenomenon (e.g., LaMassa et al. 2015; MacLeod et al. 2016, 2019; Ruan et al. 2016; Runnoe et al. 2016; Gezari et al. 2017; Sheng et al. 2017; Wang et al. 2018; Yang et al. 2018; Yan et al. 2019). The multiwavelength light curves of the two newly identified CL-AGNs are plotted in Figure 5. Both objects show a continual decrease of the mid-infrared brightness detected by the Wide-field Infrared Survey Explorer (WISE and NEOWISE-R; Wright et al. 2010; Mainzer et al. 2014) from 2010 to 2017, i.e., within the two spectroscopic epochs . The fading of the MIR emission supports the fact that the identified "turn-off" type transitions in the two objects are more likely due to the decrease of accretion rate rather than obscuring, because the MIR emission is mainly resulted from AGN-heated hot-dust which is less sensitive to dust obscuring (e.g., Sheng et al. 2017; Stern et al. 2018). MacLeod et al. (2019) recently argued that the CLQs are the extreme tail of regular quasar variability (e.g., Rumbaugh et al. 2018). The argument is based on the fact that the CLQs are found to have relatively low L/L_Edd, which is consistent with the well-known anticorrelation between L/L_Edd and variability amplitude previously revealed in quasars (e.g., Wilhite et al. 2008; Mao et al. 2009; MacLeod et al. 2010).

Zoom In Zoom Out Reset image size

We compile a sample of 26 previously identified CL-AGNs from the literature by requiring the detailed measurements of both their on and off states.⁶ Figure 6 shows the distributions of the CL-AGNs on the L_bol versus M_BH (the left panel) and L_bol versus L/L_Edd (the right panel) diagrams. The on and off states are plotted in the diagrams by the open and solid squares for each CL-AGN, respectively. The measurements given in Shen et al. (2011) and Chen et al. (2018) are adopted for the on state for each of the objects. For the off state, the value of L_bol is obtained from the corresponding value at the on state by a scaling factor determined through the change of the broad Hα line flux. The comparison samples used in the diagrams include (1) the SDSS DR7 quasars with z < 0.5 (Shen et al. 2011); (2) the SDSS DR3 narrow-line Seyfert 1 galaxies given in Zhou et al. (2006); (3) the Swift/BAT AGN sample with a spectral type classification in Winter et al. (2012); and (4) the SDSS intermediate-type Seyfert galaxies studied in Wang (2015).

Zoom In Zoom Out Reset image size

Two facts can be learned from the comparison shown in Figure 6. On the one hand, as being consistent with MacLeod et al. (2019), one can see from the figures that the CL-AGNs at on state are biased toward both low L_bol and low L/L_Edd. This bias in fact motivates MacLeod et al. (2019) to argue that the disk–wind BLR models proposed in Elitzur & Ho (2009) and Nicastro (2000) are plausible for understanding the (dis)appearance of the broad emission lines observed in the CL phenomenon, although a critical value of L/L_Edd ∼ 10⁻³ is required for the (dis)appearance when the fiducal values of a set of parameters of the disk are adopted. On the other hand, there is an overlap between the intermediate-type AGNs and the CL-AGNs at their off state. By adopting the change of accretion rate as the physical origin of the CL phenomenon, this overlap strongly implies that the so-called intermediate-type AGNs are composed of two populations. One is due to the well-accepted orientation effect, and the another the intrinsic change of accretion rate, even though the physical origin of the change is still unclear at the current stage. In fact, the analysis of the X-ray spectra in Winter et al. (2012) suggests that the Seyfert-1.5 galaxies statistically show higher neutral column densities than the Seyfert 1 galaxies, which agrees with the expectation of the unified model (e.g., Antonucci 1993). The statistics in Figure 6 suggests that the two populations might differ from each other in L_bol.

The expected timescale is still an open issue in the scenario of change of SMBH's accretion rate (i.e., the viscosity crisis, e.g., Lawrence 2018 and references therein). The viscous timescale of a viscous radial inflow is expected to be (e.g., Shakura & Sunyaev 1973; Krolik 1999; LaMassa et al. 2015; Gezari et al. 2017)

$$\begin{eqnarray}\begin{array}{rcl}{t}_{\mathrm{infl}} & = & 6.5{\left(\displaystyle \frac{\alpha }{0.1}\right)}^{-1}{\left(\displaystyle \frac{L/{L}_{\mathrm{Edd}}}{0.1}\right)}^{-2}{\left(\displaystyle \frac{\eta }{0.1}\right)}^{2}\\ & & \times \,{\left(\displaystyle \frac{r}{10{r}_{g}}\right)}^{7/2}\left(\displaystyle \frac{{M}_{\mathrm{BH}}}{1\times {10}^{8}\,{M}_{\odot }}\right)\ \mathrm{yr},\end{array}\end{eqnarray} \tag{ 4 }$$

where α is the "viscosity" parameter, η is the efficiency of converting potential energy to radiation, and r_g is the gravitational radius in unit of ${GM}/{c}^{2}$. Adopting $\alpha \,=\eta =L/{L}_{\mathrm{Edd}}=0.1$ and ${M}_{\mathrm{BH}}=1\times {10}^{8}\,{M}_{\odot }$ yields a viscous timescale being comparable to the observations for the near-ultraviolet emission radiated from inner accretion disk with r ∼ 10r_g. The timescale, however, dramatically increase to ∼10³ yr for the optical emission coming from the outer disk with $r\sim 50\mbox{--}100{r}_{g}$.

Some solutions have been proposed to alleviate the timescale crisis. One is to involve the local disk thermal instability. The evolutionary α-disk model developed in Siemiginowska et al. (1996) predicts a thermal timescale of

$$\begin{eqnarray}&&{t}_{\mathrm{th}}=2.7{\left(\displaystyle \frac{\alpha }{0.1}\right)}^{-1}{\left(\displaystyle \frac{r}{{10}^{16}\mathrm{cm}}\right)}^{3/2}{\left(\displaystyle \frac{{M}_{\mathrm{BH}}}{{10}^{8}{M}_{\odot }}\right)}^{1/2}\ \mathrm{yr}.\end{eqnarray} \tag{ 5 }$$

Husemann et al. (2016), in fact, reveal a temperature variation in Mrk 1018. The simulation carried out by Jiang et al. (2016) suggests that the development of the disk thermal instability favors low-metallicity gas. An alternative solution is to involve an accretion disk elevated by a magnetic field, which can results in a shorter variability timescale and can explain the CL phenomenon by an abrupt variation in magnetic torque (e.g., Ross et al. 2018; Stern et al. 2018; Dexter & Begelman 2019).

The CL phenomenon is far from being understood at the current stage, partially because of the limited CL-AGN sample size. Both repeat imaging and spectroscopy are necessary for expanding the sample size. Although some ongoing and forthcoming optical survey programs, e.g., the Pan-STARRS1 survey (Chambers et al. 2016), the Zwicky Transient Facility (Kulkarni 2018), the All-Sky Automated Survey for Supernovae (Shappee et al. 2014a), and the Large Synoptic Survey Telescope project (LSST Science Collaboration 2017), can provide many interesting targets for follow-up spectroscopy. How to flag the CL phenomenon efficiently by excluding the contamination caused by the AGN's normal variation in optical bands is an open issue. This issue can be fairly addressed by some recently proposed space-based ultraviolet (UV) patrol missions (e.g., Sagiv et al. 2014; Mathew et al. 2018; Wang et al. 2019), because a significant variation in the AGN's UV continuum is expected in the CL phenomenon. Additionally, a better understanding of the CL phenomenon can stem from the forthcoming larger time-domain spectroscopic surveys, for example the SDSS-V survey (Kollmeier et al. 2017).

The authors thank the anonymous referee for the careful review and helpful suggestions that improved the manuscript. J.W. and D.W.X. are supported by the National Natural Science Foundation of China under grant 11773036. This study was supported by the National Basic Research Program of China (grant 2014CB845800), by the Natural Science Foundation of Guangxi (2018GXNSFGA281007), the Bagui Young Scholars Program, and by the Strategic Pioneer Program on Space Science, Chinese Academy of Sciences (grant Nos. XDA15052600 and XDA15016500). This work was partially supported by the Open Project Program of the Key Laboratory of Optical Astronomy, NAOC, CAS. This study uses the SDSS archive data created and distributed by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. Special thanks go to the staff at Xinglong Observatory for their instrumental and observational help, and to the allocated observers who allowed us to finish the observations in ToO mode.

Facility: Xinglong 2.16 m telescope. -

Software: IRAF (Tody 1986, Tody 1993).