iPTF Discovery of the Rapid "Turn-on" of a Luminous Quasar

The source SDSS J155440.25+362952.0 was imaged by SDSS on UT 2003 April 29 (all days hereafter are in the UT system) and morphologically classified as a galaxy with $r=18.18\pm 0.01$ mag. It was targeted in the SDSS spectroscopic legacy survey as a ugri-selected quasar, selected for lying more than 4σ from the stellar locus (Richards et al. 2002) at high Galactic latitude (QSO_CAP). It turns out that while the colors measured in the SDSS survey are outside the stellar locus, with $u-g=+0.44\pm 0.04$ mag, $g-r=+0.92\pm 0.01$ mag, and $r-i=+0.48\pm 0.01$ mag, given the known redshift of the galaxy, they are inconsistent with the quasar color–redshift relation measured for the SDSS spectroscopic sample (Schneider et al. 2007).

The SDSS legacy spectrum, obtained on 2004 June 16, was determined by the spectroscopic pipeline to have a z = 0.2368 and a spectroscopic classification of a "GALAXY," with a stellar velocity dispersion of ${\sigma }_{\star }=176\pm 14$ km s⁻¹. The galaxy classification was due to the presence of strong galaxy absorption features (Ca H & K, G band, Mg i, Na D), with only weak [O iii] and [N ii] emission lines detected. According to the ${M}_{\mathrm{BH}}-{\sigma }_{\star }$ scaling relation, this velocity dispersion corresponds to a central black hole mass of $({1}_{-0.7}^{+2})\times {10}^{8}\,{M}_{\odot }$ (McConnell & Ma 2013).

We find no evidence for significant flux variations between the SDSS image in 2003 and legacy spectrum in 2004. The synthetic r-band magnitude of the 2004 SDSS spectrum (measured by projecting the best-fit spectral template onto the r-band filter) is spectroSynFlux_r = 18.82 ± 0.02 mag. The corresponding fiber magnitude for the SDSS imaging in 2003 (measured with an aperture equal to the 3'' diameter spectroscopic fiber) is r = 18.95 mag. Since the SDSS spectrum in 2004 is dominated by host-galaxy starlight in the wavelength range of the r band (${\lambda }_{\mathrm{eff}}=6231$ Å), we conclude that the SDSS image in 2003 is also dominated by host-galaxy starlight.

The source was observed by the GALEX All-Sky Imaging Survey on 2004 May 15 with a 6'' (4 pixel) radius aperture magnitude corrected for the total energy enclosed (Morrissey et al. 2007) of $\mathrm{NUV}=21.61\pm 0.35$ mag, a 5σ point-source upper limit of $\mathrm{FUV}\gt 20.6$ mag for ${t}_{\exp }=112\,{\rm{s}}$, and a background of $2.6\times {10}^{-3}$ counts s⁻¹ pixel⁻¹. Note that the color measured by GALEX and SDSS of $\mathrm{NUV}-r=+3.4\pm 0.35$ mag is entirely consistent with normal galaxies with a similar luminosity on the blue sequence (Wyder et al. 2007).

iPTF 16bco was discovered as a transient detection by the Palomar 48-in telescope (P48) on 2016 June 1 during an (intermediate) Palomar Transient Factory (iPTF) g+r-band experiment by the real-time difference-imaging pipeline run at LBNL (Cao et al. 2016), with g = 19.4 mag and r = 19.6 mag. The source was flagged as a "nuclear" transient given the measured offset of 044 from its host galaxy, within our centroiding accuracy of 08. The data were re-reduced with the PTFIDE pipeline run at IPAC (Masci et al. 2016).

On the next day after the discovery (2016 June 2), the transient iPTF 16bco was followed up with the robotic low-resolution ($R=\lambda /{\rm{\Delta }}\lambda \sim 100$) Integral Field Unit (IFU) spectrograph, part of the Spectral Energy Distribution Machine (SEDM) instrument on the Palomar 60-in telescope (P60). The spectrum was strikingly different from its archival SDSS spectrum, and three more epochs of spectroscopy were obtained with DEIMOS (Faber et al. 2003) on the Keck-II telescope on 2016 June 4 and the DeVeny spectrograph on the Discovery Channel Telescope (DCT) on 2016 June 13 and 2016 July 9.

The SEDM IFU obtained 2700 s exposures on 2016 June 2 and July 24, and data reduction was performed by the SEDM pipeline.¹⁶ Sky subtraction was performed using "A–B" extraction: two exposures of the same length are taken offset by a few arcseconds, so that the object of interest lies on an empty sky region on the next exposure, and then subtracted from each other to remove the sky lines. The spectrum is extracted in each exposure separately, and then the fluxes are summed to get the final spectrum.

The Keck spectrum was obtained with a $0\mathop{.}\limits^{\prime\prime }8$ slit, and the exposure time was 240 s. Data were reduced with usual procedures from the DEEP2 pipeline in IDL (Cooper et al. 2012; Newman et al. 2013) and in PyRAF. The flux calibration and telluric correction were done with the flux standard star BD 262606. The spectrum covers wavelengths ranging from 4550 to 9550 Å with a spectral resolution of ∼4 Å. The DCT spectra were obtained with a $1\mathop{.}\limits^{\prime\prime }5$ slit and an exposure time of 600 s and 1200 s, respectively, with a spectral resolution of ∼9 Å. Data were reduced in standard IRAF routines, and flux calibration was performed with the flux standard star BD +40d4032.

Figure 1 shows the series of follow-up spectra, which demonstrate strong, broad, Balmer emission lines characteristic of a type 1 quasar at z = 0.2368.

Zoom In Zoom Out Reset image size

We also monitored the source in three filters ($g,r,i$) with the SEDM on the P60 telescope. The data were host-subtracted using FPipe (Fremling et al. 2016). One epoch of the P60 data points on MJD 57,576 is from the GRBCam. We do not plot the i-band GRBCam data from this night, due to the large difference in the shape of its filter transmission curve in this band. The light curve of the transient iPTF 16bco is presented in Figure 2, and the photometry is given in Table 1. We adjust the g-band P48 photometry by +0.25 mag in order to match the P60 photometry. This offset is attributed to the difference in filter curves in the g band and the strong blue continuum. Note that since its discovery by iPTF on 2016 June 1 in the g and r bands, iPTF 16bco has retained a blue color, $g-r\approx -0.1$ mag, and demonstrated a rise in brightness of ∼0.5 mag over a timescale of ∼1 month.

Zoom In Zoom Out Reset image size

The source was observed in the PTF survey in 2009–2012, and no variability was detected in this time frame, with a median transient point-source upper limit over 151 epochs of $r\lt 20.9$ mag, with the last nondetection on 2012 May 28. When adding in the host-galaxy flux measured by SDSS, this corresponds to a total magnitude of $r\lt 18.10$ mag, or a ${\rm{\Delta }}r\lt 0.08$ mag during this time period. The P48 observations constrain the onset of the nuclear transient to be after the last nondetection on 2012 May 28, 4 yr before the iPTF discovery.

The source was observed with our Swift Key Project program for UV follow-up of iPTF nuclear transients (PI: Gezari) on 2016 June 21 and July 5. We extracted the source from a $5\buildrel{\prime\prime}\over{.} 0$ region with a background region of $20\buildrel{\prime\prime}\over{.} 0$ radius using the task uvotsource in HEASoft, which includes a correction for the enclosed energy in the aperture. The source was observed in the uvw2 filter with ∼1 ks exposures and detected with 19.30 ± 0.05 mag and 19.11 ± 0.05 mag in the AB system, respectively. The corresponding UV−optical color of iPTF 16bco (with a negligible contribution of UV flux from the host) is $\mathrm{NUV}-r\sim -0.5$ mag, notably bluer than the near-UV − r colors of low-redshift quasars measured by GALEX and SDSS ($\mathrm{NUV}-r=0.0\mbox{--}0.5$ mag; Bianchi et al. 2005; Agüeros et al. 2005).

Simultaneous Swift XRT observations were processed with the UK Swift Data Science Centre¹⁷ pipeline that takes into account dead columns and vignetting to extract counts from the source in the energy range of 0.3–10 keV. The X-ray count rate on June 21 and July 5 is 0.025 ± 0.0057 and 0.013 ± 0.0059 counts s⁻¹, respectively. We further obtained a 1.7 ks Swift XRT exposure on 2016 October 21, and the source was detected with 0.027 ± 0.004 counts s⁻¹. This confirms the lack of significant X-ray variability between the Swift observations. Furthermore, the combined spectrum of all the data can be modeled with an absorbed power law with a spectral index of ${\rm{\Gamma }}=2.1\pm 0.5$ and N_H fixed at the Galactic value of $1.63\times {10}^{20}$ cm⁻² (Dickey & Lockman 1990). The average unabsorbed 0.2−10 keV flux of the source is 9 × 10⁻¹³ erg s⁻¹ cm⁻², corresponding to a luminosity of $1.5\times {10}^{44}$ erg s⁻¹ by assuming an absorbed power law with ${N}_{{\rm{H}}}=1.63\times {10}^{20}$ cm⁻² and ${\rm{\Gamma }}=2.1$ at z = 0.2368.

The ROSAT upper limit in the 0.1–2.4 keV band from the All-Sky Survey in 1990–1991 (Voges et al. 1999) is 0.1 counts s⁻¹, which corresponds to an unabsorbed flux of $\sim 7\times {10}^{-13}$ erg s⁻¹ cm⁻². There are even more constraining 2σ upper limits from the XMM Slew Survey¹⁸ of $\lt 0.601$ and $\lt 0.817$ counts s⁻¹ in the 0.2–12.0 keV band on 2011 February 27 and 2015 February 08, corresponding to $\lt 2.7\times {10}^{-13}$ and $\lt 3.2\times {10}^{-13}$ erg s⁻¹ cm⁻², respectively. The latest XMM Slew Survey upper limit implies a factor of $\gt 3$ increase in flux in the Swift XRT detection on a timescale of <1.1 yr in the quasar rest frame.

We observed iPTF 16bco with the AMI-LA at 15.5 GHz on 2016 October 16.63. The source is not detected, with the 3σ upper limit of 68 μJy. We can also convert this to a 1.4 GHz upper limit of 370 μJy using a spectral index of −0.7. This is consistent with the nondetection in the VLA FIRST survey (Becker et al. 1995) from 1999, which gives an independent 1.4 GHz upper limit of 500 μJy.

The archival SDSS spectrum from 2004 was fitted by the automated spectroscopic pipeline (Bolton et al. 2012) with a combination of stellar, galaxy, and quasar templates plus emission lines. After visual inspection of the pipeline fit, we found a poor fit to the the Hα+[N ii] and [O iii] emission line complexes and refitted the spectrum with host-galaxy template and emission-line gas components using ppxf (Cappellari & Emsellem 2004; Cappellari 2016), which uses the MILES stellar template library (Vazdekis et al. 2010). The emission-line fits are shown in Figure 3 and are fitted with a narrow Gaussian with a $\sigma =420$ km s⁻¹, with no evidence for broad Hα or Hβ components. The narrow-line ratios of $\mathrm{log}$([O iii]$\lambda 5007$/H $\beta )=0.74\pm 0.13$ and $\mathrm{log}$([N ii] $\lambda 6583/$ H$\alpha )=0.47\pm 0.07$, together with L([O iii] $\lambda 5007)\,=(1.0\pm 0.1)\times {10}^{41}$ erg s⁻¹, classify the SDSS spectrum as a type 2 AGN in the LINER region (Kewley et al. 2006) in the diagnostic narrow-line diagrams (Baldwin et al. 1981; Veilleux & Osterbrock 1987; Kauffmann et al. 2003) shown in Figure 4. Note that the archival WISE colors from the all-sky survey in 2010 (Cutri et al. 2012) of W1 − W2 = 0.48 ± 0.04 mag and W2 − W3 = 1.6 ± 0.2 mag, where W1, W2, and W3 are 3.4, 4.6, and 12 μm, respectively, place the host in the region of Seyfert and star-forming galaxies (Yan et al. 2013). However, in the WHAN diagram (shown in Figure 4) for emission-line galaxies (Cid Fernandes et al. 2010, 2011), the weak equivalent width of Hα (${W}_{{\rm{H}}\alpha }=1.6\pm 0.3$ Å) classifies the galaxy as a "retired galaxy" powered by hot, low-mass, evolved (post-asymptotic giant branch) stars and not an AGN.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The transient iPTF 16bco shows two remarkable changes: a factor of 10 increase in UV flux, and a transformation from a LINER galaxy to a luminous type 1 quasar. Figure 5 shows the dramatic change of state between the SDSS spectrum in 2004 and the follow-up Keck spectrum in 2016.

Zoom In Zoom Out Reset image size

3.2.1. Continuum Variability

3.2.2. Spectral Variability

3.3.1. Variable Obscuration or Microlensing

3.3.2. Tidal Disruption Event (TDE)

We now investigate the scenario that iPTF 16bco was the result of a change of state in a preexisting quasar accretion disk. Interestingly, however, iPTF 16bco puts stringent limits on the timescale over which such accretion rate changes must occur. In the rest frame, iPTF 16bco demonstrates a dramatic change in continuum flux over a timescale of ${\rm{\Delta }}t\lt 4/(1+z)\,=3.23\,\mathrm{yr}$, or in <1.1 yr based on the archival X-ray upper limits. The timescale by which an accretion disk can change its accretion rate should be determined by the viscous radial inflow timescale, ${t}_{\mathrm{infl}}$. Furthermore, this timescale is expected to be longer for a quasar "turning on" instead of "turning off," since it scales as ${\lambda }_{\mathrm{Edd}}^{-2}$ (LaMassa et al. 2015). The ${t}_{\mathrm{infl}}$ corresponding to the Eddington ratio and black hole mass estimated for iPTF 16bco in its dim state, and assuming a radiation-pressure-dominated inner region of a Shakura–Sunyaev disk, is,

$$\begin{eqnarray*}&&{t}_{\mathrm{infl}}=1300\,\mathrm{yr}\,{\left[\displaystyle \frac{\alpha }{0.1}\right]}^{-1}{\left[\displaystyle \frac{{\lambda }_{\mathrm{Edd}}}{0.005}\right]}^{-2}{\left[\displaystyle \frac{\eta }{0.1}\right]}^{2}{\left[\displaystyle \frac{r}{10{r}_{{\rm{g}}}}\right]}^{7/2}\left[\displaystyle \frac{{M}_{8}}{2.0}\right],\end{eqnarray*}$$

a much longer timescale than the observed rapid change in continuum flux in iPTF 16bco.

One possibility could be an accretion disk eruption as the result of thermal-viscous instabilities in a partial ionization zone, analogous to the outbursts observed in cataclysmic variables (CVs) and X-ray novae (Siemiginowska et al. 1996). However, while such instabilities can produce amplitudes of a factor of 10⁴, the expected durations scale with the central mass, and so while CVs and X-ray novae show large-amplitude outbursts on the timescale of weeks to months, for an accretion disk around a ${10}^{8}\,{M}_{\odot }$ black hole, this corresponds to timescales of ∼10⁵ yr. However, state changes on ∼minute timescales have been observed in some X-ray binaries (Fender 2001; Fender & Belloni 2004), which would imply similar changes in an SMBH on the timescale of only ∼10 yr.

In the accretion disk instability model, the more narrow the unstable zone, the more rapid the timescale of variability. The size of the unstable zone depends on the accretion rate: the lower the accretion rate, the closer in the unstable region is to the inner edge of the disk. However, the smaller the unstable zone, the smaller the expected amplitude of variability. For models that demonstrate a factor of ∼10 variability ($\alpha =0.1,\dot{M}=3\times {10}^{-5}\,{M}_{\odot }$ yr⁻¹) in Siemiginowska et al. (1996), they have a "turn-on" timescale of ∼10³ yr. This is still 3 orders of magnitude longer than we require for the "turn-on" timescale of iPTF 16bco.

Interestingly, the thermal timescale itself, ${t}_{\mathrm{th}}\sim 1/\alpha {{\rm{\Omega }}}_{{\rm{K}}}$, is much shorter, adopting Equation (8) from (Siemiginowska et al. 1996):

$$\begin{eqnarray*}&&{t}_{\mathrm{th}}\sim 2.7{(\alpha /0.1)}^{-1}{M}_{8}^{-0.5}{(r/{10}^{16}\mathrm{cm})}^{1.5}\,\mathrm{yr}.\end{eqnarray*}$$

A disk with local thermal fluctuations, potentially driven by the magnetorotational instability, could be consistent with the rapid timescale of the continuum variability in iPTF 16bco (Dexter & Agol 2011). While an inhomogeneous disk has been demonstrated to fit composite difference spectra (Ruan et al. 2014) and color variability of quasars (Schmidt et al. 2012), Hung et al. (2016) find a simple disk model adequate to fit difference-flux UV/optical spectral energy distributions of individual quasars. Moreover, Kokubo (2015) argues that the tight interband correlations observed in SDSS quasar light curves are inconsistent with the inhomogeneous disk model. However, whether or not these thermal fluctuations can be coherent enough to produce a large-amplitude outburst is still to be determined.

An instability that arises in a radiation-pressure-dominated disk (Lightman & Eardley 1974) has been used to model recurrent flares in accreting Galactic X-ray binaries (Belloni et al. 1997) and has recently been applied by Saxton et al. (2015) and Grupe et al. (2015) to explain the large-amplitude soft X-ray flares in Seyfert galaxies NGC 3599 and IC 3599, respectively. In this scenario, when the internal radiation pressure of the disk become greater than the gas pressure, a heating wave propagates through the disk. This results in an enhanced local viscosity, scale height, and accretion rate, which rapidly drains the disk. The instability recurs when the inner disk fills back in. The rise time and recurrence timescale can be as short as a year and hundreds of years for a ${10}^{8}\,{M}_{\odot }$ black hole, respectively.

Another mechanism for driving large-amplitude variability in a quasar accretion disk could be related to the presence of a binary SMBH. Hydrodynamical simulations show that in a circumbinary disk, streams penetrate the disk cavity to feed the primary and secondary black hole at a periodic rate, and that at close to equal mass ratios, the perturbed circumbinary disk has an enhanced accretion rate that can be quite bursty on a timescale of ∼5 times the orbital period of the SMBH binary (Farris et al. 2014). Such a scenario could in principle be testable, due to the intrinsically periodic nature of the outbursts. Furthermore, one could look for periodic changes in the broad-line profiles, if they originate in the circumbinary disk. However, the circumbinary disk outbursts in Farris et al. (2014) have a sawtooth pattern, with a rise time much shorter than the decay time. Already the light curve of iPTF 16bco is not in good agreement with this model, given its rebrightenning by ∼0.5 mag during its high state (see Figure 2).

The continued intrinsic variability during the high state of iPTF 16bco, as also seen in the study of changing-look quasars by MacLeod et al. (2016), may also provide insight as to the nature of what caused the "changing look" of the quasar. The optical variability amplitudes of these sources in their type 1 quasar states of ∼0.5–1.0 mag are on the high-amplitude tail of what is typically observed on these timescales for quasars of a similar luminosity range (MacLeod et al. 2012), although we note that in both of our studies the changing-look quasars were selected by their optical variability. However, intrinsic variability is also observed in the changing-look quasar presented in LaMassa et al. (2015), which was selected based on its spectral changes. In fact, the rapid rise and power-law decline of its light curve were interpreted as a signature of a TDE (Merloni et al. 2015). However, the light curves of iPTF 16bco and the other changing-look quasars with archival photometry have large fluctuations that are inconsistent with the smooth, power-law decline observed in the optical light curves of known TDEs (e.g., Gezari et al. 2008; van Velzen et al. 2011; Gezari et al. 2012; Arcavi et al. 2014; Holoien et al. 2014, 2016a, 2016b). The erratic intraburst behavior in the two-state (hot and cold) α instability models of Siemiginowska et al. (1996) could be promising. A potential testable prediction is that in these models, the accretion disks spend the majority of their time in the low state. The variability during the enhanced accretion state in iPTF 16bco could also be a signature of clumpy accretion in an advection-dominated accretion flow (ADAF), for which cold clumps form in the accretion flow owing to instabilities in the radiation-dominated regions of the disk (Wang et al. 2012)

The dramatic change in accretion rate from ${\lambda }_{\mathrm{Edd}}\lesssim 0.005$ to ${\lambda }_{\mathrm{Edd}}\sim 0.05$ inferred for iPTF 16bco could be accompanied by a structural change in the accretion flow if the quasar accretion disk is transitioning from a radiatively inefficient to radiatively efficient mode. Note that such changes in X-ray binaries are often accompanied by changes in jet activity. The implied high-state radio-to-optical flux density ratio for iPTF 16bco of $R=\mathrm{log}({S}_{1.4\mathrm{GHz}}/{S}_{\mathrm{opt}})\lt 0.8$ and radio luminosity ${L}_{{\rm{R}}}\lt 3\times {10}^{22}$ W Hz⁻¹ are typical of radio-quiet AGNs (e.g., Padovani et al. 2011). While these values are consistent with the LINER classification from the optical spectrum, it is surprising that if the accretion event in iPTF 16bco were triggered by a disk instability, there would be no evidence for a jet or outflow during its high state in the radio. This is in contrast with X-ray binaries and CVs, which generally show flaring at radio, optical, and X-ray wavelengths alongside strong Balmer emission lines. Similarly, the fundamental plane of black hole activity (e.g., Plotkin et al. 2012; Saikia et al. 2015) predicts ${L}_{{\rm{R}}}$ significantly greater than $\sim {10}^{22}$ W Hz⁻¹ when ${L}_{{\rm{X}}}=1.5\times {10}^{44}$ erg s⁻¹.

iPTF 16bco is one of only a dozen other changing-look quasars (here defined as ${M}_{i}\lt -22$ mag, L([O iii]) > 10⁴¹ erg s⁻¹), roughly half of which have been caught in the act of "turning on" by demonstrating the sudden appearance of broad lines. Figure 6 shows the redshift and [O iii] luminosity of all the changing-look quasars in the literature that pass our [O iii] luminosity cut (thus, we exclude SDSS J0126-0839 and SDSS J2336+0017 from Ruan et al. 2016), color-coded by whether they show appearing broad lines or disappearing broad lines. We also do not include three changing-look quasars from the MacLeod et al. (2016) sample that do not have good coverage of the broad Hα line in its high state (appearing SDSS J214613 at z = 0.62, disappearing SDSS J022562 at z = 0.63, and both appearing and disappearing SDSS J022556 at z = 0.50). Given that all the other appearing changing-look quasars are from MacLeod et al. (2016), there appears to be a bias toward finding appearing broad lines in higher-redshift galaxies. This is likely due to the fact that the BOSS spectra extend to longer wavelengths than the SDSS spectra, and thus finding an appearing broad Hα line in the BOSS spectrum is more likely at higher redshift than finding a disappearing broad Hα line in the SDSS spectrum.

Zoom In Zoom Out Reset image size

Figure 6 also shows the [O iii] $\lambda 5007$ versus broad Hα luminosity in all the changing-look quasars in their type 1 state, in comparison to the full SDSS quasar sample. We determine the [O iii] $\lambda 5007$ luminosity from the line flux measured in the SDSS DR7 SpecLine table and the DR12 interactive spectrum line measurement table. We use the broad Hα fluxes from the Shen et al. (2011) catalog for broad-line quasars, or from the literature when available. All luminosities are calculated for our adopted cosmology. Note that iPTF 16bco is on the edge of the normal quasar distribution, while the other changing-look quasars reported in the literature appear to lie squarely in the distribution of normal quasars in this parameter space.

The enhanced broad Hα luminosity observed in iPTF 16bco relative to [O iii] in comparison to normal quasars, as well as the previously discovered changing-look quasars, is likely a signature of its rapid transition to a type 1 state. As discussed in Section 3.2.2, given the extended size of the narrow-line region, the [O iii] line will lag in its response to a continuum flare in comparison to the broad emission lines due to light-travel time effects. Interestingly, in the MacLeod et al. (2016) sample, the rise time in the continuum flux of the quasars was serendipitously measured by photometric monitoring to be ∼1000 days. Given the significantly weaker [O iii] line emission relative to broad Hα in iPTF 16bco compared to these objects, this would imply an even shorter "turn-on" timescale, in agreement with the inferred "turn-on" timescale for the continuum in iPTF 16bco of $\lesssim 1\,\mathrm{yr}$.

One interesting aspect of changing-look quasars is the lack of strong changes in Mg ii line emission, despite the dramatic changes in the Balmer lines. This was explained by MacLeod et al. (2016) as being due to the relatively weak responsivity of the Mg ii line to continuum flux changes, as has been measured in rest-frame UV reverberation mapping studies (Cackett et al. 2015). Given that Mg ii is a low-ionization line, its weak responsivity can be explained as a consequence of the stratification of the broad-line region (Korista & Goad 2004). Unfortunately, we do not have short-enough wavelength coverage in the archival or follow-up spectra to determine the presence and/or response of the Mg ii line during the change of state in iPTF 16bco.