A Mystery in Chamaeleon: Serendipitous Discovery of a Galactic Symbiotic Nova

CN Cha was first identified as a Mira variable in 1963 and was observed by multiple surveys throughout the 20th century that took measurements consistent with this initial conclusion. A visual summary of the photometric data since 2000 is given in Figure 2. We gather multi-epoch photometry spanning nearly 20 yr ranging from the near-UV to the far-infrared. The information summarized in Figure 2 indicates that CN Cha experienced an outburst event at some point in 2013. We find that the pre-outburst SED is best described by the sum of two blackbodies, one cool (2000 K) bright component, and one hot (10,000 K) dim component. Observing the SEDs from before and after this event, it is clear that the object became much brighter in almost all bands for which we can make a reasonable comparison. This brightness difference is most extreme in the UV, where an observed increase in flux by a factor of 10,000 is seen between UVOT and SkyMapper measurements. It should be noted that these SkyMapper measurements in the UV were taken very soon after the outburst event, in 2014 March, so it is plausible to assume that the SED now is somewhat different than it was then, especially given the dimming that was observed later in the ASAS–SN light curve.

Zoom In Zoom Out Reset image size

Finally, we show the TESS light curve in Figure 3, which we classify as "archival" despite the fact that it was taken after our own observation. The main conclusions from the TESS data are that (1) they indicate that the star is still steadily dimming at a rate consistent with the ASAS–SN observations and (2) they suggest the star has short-term variability on the timescale of several hours.

Zoom In Zoom Out Reset image size

The European Space Agency's astrometric space mission Gaia, observed CN Cha during its first window of observation used in the mission's second data release (DR2; DR2 Gaia Collaboration et al. 2016, 2018). This observation window was from 2014 July 24 to 2016 May 23. The mission observed the star to be peculiar in a number of ways that alerted the authors that the star may be interesting. The key observations that piqued our interest were:

• 1.  
  The photometric observations gave G = 7.41, G_BP = 7.72, G_RP = 6.98 mag, making the star relatively bright.
• 2.  
  The star was assigned a parallax of

  $$\begin{eqnarray}&&\varpi =0.3142\pm 0.0249\,\mathrm{mas}\end{eqnarray} \tag{ 1 }$$

  putting the star's distance at ${3.18}_{-0.23}^{+0.27}$ kpc, implying an absolute G band magnitude of −5.1 mag, making it unusually luminous.
• 3.  
  The astrometric pipeline assigned zero astrometric excess noise (AEN) to this object, which indicates that the astrometry does not show obvious problems. The Renormalized Unit Weight Error (RUWE) is 1.19.
• 4.  
  The star was not classified as variable.

The first three of the above items were enough to identify this object as one of interest for follow-up.

CN Cha was first identified as a Mira variable star in 1963 as reported by Hoffmeister (1963). At that time, the star was found to vary between magnitude 15 and 17,¹³ in the Johnson I band.

The star was later identified by the Infrared Astronomical Satellite (IRAS) as a strong point source in all four bands (12, 25, 60, and 100 μm) as well as being given an 82% probability of being variable based on analysis of the 12 and 25 μm flux densities and their uncertainties (Neugebauer et al. 1984).

Despite being very bright in the optical at the current epoch, it was not recorded as a source in any of the catalogs created from observations performed by the Hipparcos satellite (ESA 1997; Høg et al. 2000). The completeness of the Tycho-2 catalog is 99% for stars brighter than 11 magnitude in the V band (Høg et al. 2000), suggesting that CN Cha was likely dimmer than 11th magnitude in the V band during Hipparcos's observations from 1989 to 1993.

The Digitized Sky Survey (DSS) consists of a series of photometric plate observations done on the Oschin and UK Schmidt Telescopes between 1983 and 2006 (Lasker et al. 1990; 1996). CN Cha was observed by the UK Schmidt Telescope using the RG610 filter (approximating the Johson–Cousins R band) on 1991 April 12th. We show this observation in the left hand panel of Figure 1. The star was reported to have magnitude RG610 = 13.26 mag in this band by the DSS (Lasker et al. 1990).

The star was also observed by the DSS in January of 1976 in a GG395 filter at a bluer wavelength and in March of 1996 in a RG715 filter at a redder wavelength. While these exposures are available online,¹⁴ we could not determine zero-point fluxes for stars in these bands or stars within the field with reference magnitudes in these bands, so we do not report any photometry for these images.

The Deep Near-Infrared Survey (DENIS) was a near-infrared survey operated out of the 1 m European Southern Observatory (ESO) telescope at La Silla, Chile (Epchtein et al. 1994). DENIS also gathered multiple observations of 355 million objects in each of its three bands: I (0.8 μm), J (1.2 μm), and K_s (2.1 μm) (Fouqué et al. 2000). The DENIS survey observed CN Cha on three occasions from early 1999 to January of 2000 (DENIS Consortium 2005). However all observations in the J and K_s bands were brighter than the saturation limits of the DENIS survey, as was the second observation in the I band (Epchtein et al. 1994). For this reason we only use the two DENIS observations in the I band that were unsaturated. The range of flux that these observations cover is indicated in Figure 2.

The Two-Micron All-Sky Survey (2MASS) was a full-sky survey in the J, H (1.65 μm), and K_s bands operated on two dedicated 1.3 m telescopes at Mount Hopkins, Arizona and Cerro Tololo, Chile between 1997 June and 2001 February. The survey was estimated to be complete at signal-to-noise greater than 10 to approximately 15th magnitude in the H band. CN Cha was observed by 2MASS on 2000 January 13th (Skrutskie et al. 2006).

The All-Sky Automated Survey (ASAS) was designed as one of the first major efforts to obtain time-variability information on a large fraction of stars in the night sky (Pojmanski 1997). The ASAS survey obtained observations of CN Cha from 2000 November 21 to 2009 November 27 (Pojmanski 2002a, 2003, 2004). The light curve for the star is reported in two separate online catalogs that we were able to find. We use both catalogs, as they seem consistent with one another, and discuss the relation between them in Appendix.

The ASAS survey used these observations to identify CN Cha as a Mira variable star with a period of 260 days. The light curves from the original ASAS survey were reanalyzed in 2016 by Vogt et al. (2016) with more rigorous analysis techniques. In this reanalysis, CN Cha has a period of 253.24 days.

The Radial Velocity Experiment (RAVE) obtained a spectrum of CN Cha on 2010 February 20 (Kordopatis et al. 2013). The RAVE team very generously provided us with a copy of the spectrum taken in 2010 in advance of its planned release in 2020. The star's spectrum shows clearly very low T_eff with broad TiO absorption features, consistent with a cool giant star. Unfortunately, the RAVE pipeline fits seem to have hit the boundary of the stellar parameter grids and therefore extracted stellar model parameters are not reliable (RAVE Team 2020, private communication). We leave the analysis of this spectrum to future work.

The AKARI satellite was a Japanese astronomical all-sky survey in six infrared bands between 9 μm and 200 μm (Murakami et al. 2007). This survey was carried out between 2006 May 6 and 2007 August 8, during which time AKARI observed CN Cha and reported measurements for its flux in band passes centered at 9 μm and 18 μm and non-detections in all longer wavelength bands at 65, 90, 140, and 160 μm (Yamamura et al. 2010).

The Neil Gehrels Swift Observatory is a space observatory mission directed by NASA and built for the main purpose of detecting Gamma Ray Bursts (GRBs).¹⁵ The Swift UV and Optical Telescope (UVOT) observes the sky in four broad bands UVW2 (1928 Å), UVM2 (2246 Å), UVW1 (2600 Å), and U (3465 Å) spanning the far to near-UV (Kuin et al. 2015). Data from UVOT was used to create the Swift/UVOT Serendipitous Source Catalog (SUVOTSSC), which catalogs a number of point sources in these bands (Page et al. 2014; Yershov 2014). CN Cha was observed six times by UVOT and we display the photometry from 2008 January in Figure 2.

The American Association of Variable Star Observers (AAVSO) Photometric All-Sky Survey (APASS) is a survey performed primarily in five bands: Johnson-Cousins B and V as well as Sloan g',r', and i' (Henden et al. 2009; Henden & Munari 2014). With their latest data release (DR10), there have been observations in the Sloan u' and z' bands as well as the Y band (Henden et al. 2018).

The epoch photometry from the APASS survey is also available through the AAVSO's International Variable Star Index (VSX).¹⁶ The VSX data includes g', r', and i' values for five epochs spanning from 2012 March 26 to 2014 April 5. The last three of these five epochs additionally have photometric measurements in the B, V, and z' bands. This photometric information was essential in constraining the behavior of CN Cha in an otherwise unobserved portion of its evolution as shown in Figure 2, where we can see its rise in optical luminosity from magnitude 16 to magnitude 8.

As we have the most information on the long-term variability of CN Cha in the V band, in order to have the largest number of epochs for comparison we use the APASS g' andr' bands to calculate an effective V band magnitude for the two earliest observations where there is no directly measured V band magnitude. The calibration we use is

$$\begin{eqnarray}&&V=g^{\prime} -0.52(g^{\prime} -r^{\prime} )-0.03\,\mathrm{mag},\end{eqnarray} \tag{ 2 }$$

which comes from Jester et al. (2005). It should be noted that this relation is derived using relatively well-behaved stars, so our use of the relation here for an object with such an abnormal SED (as shown in Figure 2) may not be ideal.

The Wide-field Infrared Survey Explorer (WISE) observed CN Cha on 2010 February 20 and 27, using about 30 individual exposures for their total photometric measurement (Cutri et al. 2012; ). CN Cha was observed as a strong point source in all four of the WISE photometric bands W1, W2, W3, and W4. More than 15% of the pixels were determined to be saturated in the W1 and W2 bands (Cutri et al. 2012), thus we quote the measured photometry as lower bounds of the true flux at these wavelengths.

Once the cryogenic cooling on the WISE satellite became non-operational, observations could no longer be performed in the W3 and W4 bands, but the mission continued to observe the infrared sky in the W1 and W2 bands as the NEOWISE mission (Mainzer et al. 2014). While the NEOWISE mission observed CN Cha on several occasions, it was highly saturated in all exposures, having derived mean magnitudes of W1 = 3.22 and W2 = 2.19 mag, which is much brighter than the objects that NEOWISE was designed to measure (Mainzer et al. 2014). Regardless, we provide a further analysis of the light curve created by NEOWISE in Section 4.3.

The SkyMapper Survey is a photometric survey of the southern sky performed on the 1.35 m telescope at the Siding Spring Observatory in Australia. The survey uses six photometric bands, which are u, v, g, r, i, and z, and they had their first data release in January of 2018 (Wolf et al. 2018). CN Cha was not included in the stellar catalog of SkyMapper DR1 despite being observed in all six bands on three separate nights in 2014 March, likely due to its saturation in many of the bands.

To make use of these data, we downloaded the images from the SkyMapper website¹⁷ and derived magnitudes based on reference to a nearby star, which is reported in the stellar catalog of SkyMapper with magnitudes in all six bands. The reference star we used has associated SkyMapper ID 285891100 and Gaia ID 5199012694392465792. We then derived fluxes for CN Cha in the SkyMapper bands by using the AB zero-point magnitudes reported in Wolf et al. (2018). CN Cha was saturated in all SkyMapper exposures in the g, r, i, and z bands, so we only report measurements for the u and v bands. We derive u = 9.08 ± 0.20 and v = 8.59 ± 0.17 mag (in AB system) for CN Cha.

The All-Sky Automated Survey for Supernovae (ASAS–SN) is a program aimed at searching for and monitoring bright supernovae in the variable sky (Shappee et al. 2014). The project currently consists of 24 telescopes that monitor a large fraction of the night sky for variable sources down to magnitudes V < 18.¹⁸ CN Cha was observed by ASAS–SN from 2016 December 28 to 2018 February 9 in the V band and was placed in the ASAS–SN variable star catalog as a Gamma Cassiopeia (GCAS) variable star (Jayasinghe et al. 2020). Based on communication with the ASAS–SN team, this classification is probably not correct, in part due to the fact that CN Cha is quite bright for the range of magnitudes that ASAS–SN observes (Kochanek et al. 2017).

It is clear from the ASAS–SN light curve, shown on the right side of the top panel of Figure 2, that CN Cha remained at an almost constant magnitude of V ∼ 8 mag for a significant portion of the ASAS–SN monitoring before beginning to dim by 2017 September. One may be worried that the constant magnitude region is simply due to saturation of the ASAS–SN observations. Again, based on communication with the ASAS–SN team, along with the fact that this measured magnitude is consistent with the G band magnitude measured by Gaia, this flat portion being due to saturation seems unlikely, but is still plausible (private communication, Tharindu Jayasinghe, ASAS–SN Team). We further analyze these data in Section 4.2.

CN Cha was observed by the Dark Energy Camera (DECam), an instrument created as part of the Dark Energy Survey (DES, Abbott et al. 2018) and mounted on the Blanco 4 m telescope at Cerro Tololo Inter-American Observatory (CTIO) in Chile. The observation was done as part of a search for substructure around the Magellanic clouds.¹⁹ One exposure in the DES r band, taken in 2016 June is shown in Figure 1. CN Cha is clearly saturated on that image, so we did not derive photometric measurements.

The Transiting Exoplanet Survey Satellite (TESS) is a photometric survey satellite launched on 2018 April 18 whose main scientific purpose is the discovery of exoplanets (Ricker et al. 2015). TESS observes single patches of the sky in 27.4 days segments, providing photometric measurements at a cadence of about 30 minutes. CN Cha was observed in TESS Sector 11 from 2019 April 22 to May 20.

To determine the light curve (LC) of the star, we have extracted the 100 × 100 pixels cutouts from the calibrated full-frame images. The star has ∼1200 observations. Since CN Cha has a nearby bright star ∼30'' (1.4 pixels) away, we decided to do a simultaneous PSF fit of the 16 × 16 pixel area around the star throughout all the epochs, while including three nearby stars in the model, and allowing for small rotation and translation between individual images. The PSF model was a linear combination of two Gaussian curves oversampled at five times the pixel size of the image, that was allowed to vary from frame to frame. From this modeling, we extract the light curve (LC) of CN Cha and of nearby stars. While the proper detrending and calibration of the LCs is beyond the scope of this paper, the extracted LCs of other stars in the field of CN Cha are quite constant with a typical median absolute deviation (MAD) of the LC of <0.5%, the MAD of the CN Cha LC is 2% and it shows an almost linear decline in flux by about 10% throughout the time of the TESS observations (27 days). That dimming corresponds to roughly a rate of 1.4 mag yr⁻¹ (see Figure 3), which is consistent with the dimming starting at approximately Julian date 2458000 in the ASAS–SN light curve (see Figure 5).

It seems safe to conclude from several points of observation in the archival data laid out above that CN Cha was a Mira long-period variable star in its recent past. This is supported by its ASAS light curve shown at the left side of the top panel of Figure 2 and in detail in Figure A1. This observation is further supported by its archival identification as a Mira variable when it was first identified (Hoffmeister 1963), its identification as a likely variable by IRAS (Neugebauer et al. 1984), the NIR colors, e.g., in 2MASS J − K = 1.69 mag (Epchtein et al. 1994; Skrutskie et al. 2006), and the RAVE spectrum gathered in early 2010 and discussed briefly in Section 2.10 (Kordopatis et al. 2013).

Here we perform a more detailed analysis of the ASAS–SN light curve that was presented in Section 2.16. In particular, we want to note the timescale of the plateau phase t_pl, which is typically used as a comparison for slow novae, and the peak absolute magnitude and decay time, which are observables typically used to compare classical novae events to one another. The latter quantities are usually quantified using the extinction-corrected, peak absolute magnitude in the V band (${M}_{V,\mathrm{peak}}^{0}$ mag) and the time it takes for the event to decay by two magnitudes from this peak (t₂). Our analysis is illustrated in Figure 5.

Zoom In Zoom Out Reset image size

For the timescale of the plateau phase, we use a rough estimate spanning from the beginning of the ASAS–SN observations (which are roughly indicated as the beginning of the plateau phase by Figure 2) and the end of the fourth contiguous epoch of ASAS–SN observations, which again is consistent with the onset of decay in the optical bands. This estimate provides t_pl = 1153 days, or just over 3 yr. We measure t₂ using a linear fit in magnitude space to the light curve for Julian Date greater than 2,458,000 days (the linear decay region). We then find the point at which this linear fit crosses two magnitudes decay from the peak observed apparent magnitude and define t₂ as the difference between that crossing time and the time that the peak magnitude was reached. We find t₂ = 807 days.

We find ${M}_{V,\mathrm{peak}}^{0}$ using the peak apparent magnitude observed in the ASAS–SN light curve (${m}_{V,\mathrm{peak}}=7.773$). We then get the peak absolute magnitude using the Gaia distance

$$\begin{eqnarray}&&{M}_{V,\mathrm{peak}}={m}_{V,\mathrm{peak}}-5{\mathrm{log}}_{10}\ \left(\displaystyle \frac{d}{10\,\mathrm{pc}}\right)=-4.74\end{eqnarray} \tag{ 3 }$$

and correcting for extinction using E(B − V) = 0.17 mag from Schlegel et al. (1998) and R_V = 3.1 this gives us

$$\begin{eqnarray}&&{M}_{V,\mathrm{peak}}^{0}={M}_{V,\mathrm{peak}}-{R}_{V}E(B-V)=-5.27.\end{eqnarray} \tag{ 4 }$$

Finally, in order to get an estimate of the total amount of energy emitted during this outburst event, we calculate the total energy in the V band emitted over the course of the ASAS–SN observations. We do this by fitting a Gaussian Process (GP) regression, available through the scikit learn python package, to the extinction-corrected ($E(B-V)=0.17$) ASAS–SN light curve (Pedregosa et al. 2011). This GP fit is shown (in observed magnitude space) in Figure 5. We then integrate the fit GP over the full course of the ASAS–SN observation. Using this integral in combination with the Gaia distance and an effective wavelength for the V band of λ_V = 5450 Å we derive the total energy in the V band of E_V = 2.8 × 10⁴⁵ ergs.

Though the observations made by the NEOWISE mission of CN Cha were saturated in all observations, we thought it useful to study the light curve if only to get a lower bound on the total amount of energy emitted in the W1 and W2 bands over the course of the NEOWISE observations. We restricted our analysis to photometric points from the NEOWISE catalog that were within 36 (0001) from CN Cha and included the original WISE observations. The NEOWISE data are grouped, with several observations occurring over a few days semi-anually (roughly 2014–2018 February and August). As these data are quite variable, we average all data points within each group to create a smoother overall light curve. We then integrate this light curve in a piecewise linear manner to get a total flux emitted over the observed period. We find 4.1 × 10⁴⁵ ergs and 5.4 × 10⁴⁵ ergs as lower bounds on the energy emitted in the W1 and W2 bands respectively. Additionally, there is no indication that the dimming that is seen in the V band is accompanied by dimming in the NEOWISE bands.

It is clear from the spectrum that the star has an abnormally large Balmer decrement. In this subsection, we will explore these lines in some detail. We first give quantitative measurements of the Balmer decrement and then use these numbers to infer the extinction needed to explain these decrements solely through dust absorption. We give phenomenological fits to the lines and the physical interpretation of these fits. Finally, we also calculate the equivalent widths of the lines.

Integrating the flux in each line and comparing we get Hα/Hβ = 90.8 (as opposed to the expected value of 3) and Hγ/Hβ ≈ 0.08 (typical value 0.45, Draine 2011). We note that the Hα emission was saturated in all exposures except for the 10 s exposure, so we use this exposure alone to measure the Hα/Hβ ratio.

This abnormal Balmer decrement would usually be indicative of extensive reddening, however, as is noted in Williams et al. (2017), nebulae commonly exhibit strong Balmer decrements, which can be due in large part to self-absorption in the higher order lines. We will come back to this in Section 4.5. As a nebula is only one of the explanations we explore below, we will nonetheless calculate the extinction that would be needed to explain the Balmer decrement that we see if there were no self-absorption. Estimating the extinction between the Hα and Hβ lines as

$$\begin{eqnarray}&&{\rm{E}}(\beta -\alpha )=2.5{\mathrm{log}}_{10}\ \left(\displaystyle \frac{1}{3}\displaystyle \frac{{F}_{{\rm{H}}\alpha ,\mathrm{obs}}}{{F}_{{\rm{H}}\beta ,\mathrm{obs}}}\right),\end{eqnarray} \tag{ 5 }$$

gives E(β − α) = 3.70 mag. Similarly, we estimate E(γ − β) = 1.86 mag. Using the R_V = 3.1 extinction curve from Weingartner & Draine (2001), we can infer that this extinction would imply a visual extinction of $E{(B-V)}_{\mathrm{Ba}}=3.07$ mag, where we have used the subscript Ba to indicate that this extinction measurement comes from the Balmer decrement. Using this extinction measurement to correct the absolute Gaia G band magnitude calculated in Section 2.2 would give an intrinsic magnitude of G₀ = −14.4 mag.

Even in the case that $E{(B-V)}_{\mathrm{Ba}}$ is an accurate estimate of the true extinction, the above correction is not necessarily accurate given the abnormal distribution of light in the G band, as indicated by the Du Pont spectrum in Figure 4. However, no matter the SED, CN Cha would be intrinsically very bright if $E{(B-V)}_{\mathrm{Ba}}$ were an accurate estimate of the extinction. For this reason, it is reasonable to conclude that either $E{(B-V)}_{\mathrm{Ba}}$ is not an accurate measure of the extinction or CN Cha is intrinsically as bright as a galaxy. While there have been transients with similar lifetimes and peak magnitudes as this, such as those found in the SPIRITS survey Kasliwal et al. (2017), we determine the former to be the more likely scenario.

In Figure 6, we show a zoom-in of the Balmer emission clearly visible in Figure 4. Note that while the lines are all displayed on the same scale, the Hα line is processed from the 10 s exposure alone (since it is saturated in all other exposures) while the other lines are simply cutouts of Figure 4. It is clear from inspecting the Balmer lines that they all exhibit strong blueshifted absorption in P-Cygni-like profiles. These profiles are typically indicative of a massive outflow of gas from the star. Each line is also very broad with widths on the order of $120\,\mathrm{km}\,{{\rm{s}}}^{-1}$ (see below).

Zoom In Zoom Out Reset image size

Attempts to fit the lines with either single Gaussian profiles or single Voigt profiles for the emission and absorption (over the wavelength regions shown in Figure 6) were unsuccessful. In an attempt to interpret these profiles shapes, we fit a five-component Gaussian model to each line. The model is defined as

$$\begin{eqnarray}&&{F}_{\mathrm{mod}}(\nu )={F}_{\mathrm{emit}}(\nu )-{F}_{\mathrm{abs}}(\nu ),\end{eqnarray} \tag{ 6 }$$

where F_mod(ν) is the model prediction as a function of frequency, F_emit(ν) is the component dedicated to capturing the emission profile and F_abs(ν) is the component meant to capture the absorption profile. We devote three Gaussian curves to the emission profile (F_emit(ν)) and two Gaussian curves to the absorption profile (F_abs(ν)). Within the emission or absorption profiles, respectively, all constituent Gaussian curves are enforced to have the same mean frequency, while all Gaussian curves are allowed to have different amplitudes and widths. The parameters of our model are then the central frequency of the emission profile (ν_e), the central frequency of the absorption profile (ν_a), the amplitudes of the Gaussian curves for the emission profile (${A}_{e,1}$,${A}_{e,2}$,${A}_{e,3}$), the widths of the Gaussian curves for the emission profile (${\sigma }_{e,1}$,${\sigma }_{e,2}$,${\sigma }_{e,3}$), the amplitudes of the Gaussian curves for the absorption profile (${A}_{a,4}$,${A}_{a,5}$), and the widths of the Gaussian curves for the absorption profile (${\sigma }_{a,4}$,σ_a,5). This leaves us with 12 free parameters.

The physically relevant parameters of these fits are given in Table 1. The radial velocity shifts RV${}_{\mathrm{emit}/\mathrm{abs}}$ are derived from the central frequencies of the fit profiles ν_a/e and the velocity Full-Width at Half-Maximum (FWHM) of the lines is derived by taking the frequency difference of the intersection of the profiles with their half-maximums. The dimensionless equivalent widths W, defined as

$$\begin{eqnarray}&&W=\int \displaystyle \frac{d\nu }{{\nu }_{0}}\left(1-\displaystyle \frac{{F}_{\nu ,\mathrm{obs}}}{{F}_{\nu ,0}}\right),\end{eqnarray} \tag{ 7 }$$

where ν₀ is taken to be the rest frame wavelength of the particular line in air, ${F}_{\nu ,\mathrm{obs}}$ is the observed flux, and ${F}_{\nu ,0}$ is the background flux that would have been observed if the line were not present. We assume ${F}_{\nu ,0}$ to be a constant and go about estimating it by taking the two Echelle orders on either side of the line of interest, sigma-clipping them to remove smaller lines that may be present, and taking the mean of the resulting set of data. This method may not be entirely accurate for the Hγ line, where the continuum is not detected at high signal-to-noise. Finally, the absorption-to-emission ratio given in Table 1 is estimated by integrating both the fitted emission and absorption profiles over the same wavelength regions used for the equivalent width estimates and taking the ratio of these integrals. We find the absorption to be between one-half and one-third of the emission across all lines.

Table 1.  The Physically Relevant Parameters of Our Five-component Gaussian Fit to the Profiles of the Balmer Lines Shown in Figure 6

Line	RVemit (km s−1)	RVabs (km s−1)	${({\rm{\Delta }}v)}_{\mathrm{FWHM},\mathrm{emit}}$ (km s−1)	${({\rm{\Delta }}v)}_{\mathrm{FWHM},\mathrm{abs}}$ (km s−1)	W	Absorption/Emission
Hα	−14.04 ± 0.46	−34.51 ± 0.19	166.74	89.03	−1.81	0.44
Hβ	−9.26 ± 1.12	−29.56 ± 0.18	121.34	72.65	−0.12	0.48
Hγ	3.19 ± 2.67	−28.25 ± 0.28	118.35	59.26	−0.02	0.33

Note. The first column specifies the line, columns 2 and 3 give the velocity blueshifts that were fit to each of the lines in emission and absorption, while columns 4 and 5 give the Full-Width at Half-Maximum (FWHM) of the profiles for the lines in both emission and absorption. Column 6 gives the equivalent width of the line and column 7 gives the ratio of absorption to emission in the line, both calculated as explained in the text.

Download table as:  ASCIITypeset image

We identify the emission lines [N ii] 5756 and [N ii] 6585, the latter of which sits on top of the wings of the Hα line. The ratio of these emission lines can be used as a temperature diagnostic (Draine 2011). After subtracting out the background by fitting polynomials of degree 2 and 6 to the regions surrounding the [N ii] 5756 and [N ii] 6585 lines respectively, we sum their fluxes and find a ratio of 2.385. If we additionally account for the extinction that occurs between these two lines based on the $E(B-V)$ from Schlegel et al. (1998) the line ratio becomes:

$$\begin{eqnarray}&&\displaystyle \frac{[{\rm{N}}\,{\rm\small{II}}]\ 5756}{[{\rm{N}}\,{\rm\small{II}}]\ 6585}=2.583.\end{eqnarray} \tag{ 8 }$$

While, as just mentioned, this ratio is usually used as a temperature diagnostic, at such large values it is more useful as a density diagnostic (Draine 2011). Assuming a gas temperature of T = 10⁴ K, this ratio is indicative of an electron density of n_e ≳ 10⁷ cm⁻³ (Draine 2011). These high densities would be consistent with self-absorption in the hydrogen recombination lines that we posited in the last section (Netzer 1975; Drake & Ulrich 1980).

Here we review the dynamics of CN Cha within the Galaxy that are derived from the astrometric measurement from Gaia Collaboration et al. (2018) and a radial velocity of ∼0 km s⁻¹, motivated by our fits to the Balmer lines given in Section 4.4. This is laid out in Figure 8.

Although CN Cha is spatially coincident on the sky with the Chamaeleon Complex, a star-forming molecular cloud complex lying between the Cha I (∼25 separation) and Cha III (∼4° separation) clouds, its distance as derived from its Gaia parallax (${3.18}_{-0.23}^{+0.27}$ kpc) places it 20× farther than the Chamaeleon Cloud (∼180 pc, Zucker et al. 2019). Furthermore, the proper motion of CN Cha (a quantity that is in general much better measured by Gaia) is also inconsistent with the proper motions of stars known to be associated with the Chamaeleon Complex. To be precise, the measured proper motion of CN Cha is $({\mu }_{\alpha * },{\mu }_{\delta })=(-5.42,7.02)\,\mathrm{mas}\,{\mathrm{yr}}^{-1}$ while the proper motions of HD 97300 ($({\mu }_{\alpha * },{\mu }_{\delta })=(-21.01,-0.61)\,\mathrm{mas}\,{\mathrm{yr}}^{-1}$) and HD 97048 ($({\mu }_{\alpha * },{\mu }_{\delta })=(-22.44,1.31)\,\mathrm{mas}\,{\mathrm{yr}}^{-1}$), both YSOs known to be associated with the Cha I cloud, are significantly different with the motion of CN Cha (Luhman 2008; Gaia Collaboration et al. 2018).

In Figure 8, we compare the past orbit of CN Cha with a star known to be in the Chamaeleon Cloud Complex (HD 97048) and the orbit of the Sun. it seems quite clear that CN Cha has dynamics that are consistent with a thick-disk star, since the maximum vertical extent of it is orbit is nearly 2 kpc above/below the disk plane and it is moving in a prograde fashion. More quantitatively, if we relate its vertical action ${J}_{z}=75.3\,\mathrm{kpc}\,\mathrm{km}\,{{\rm{s}}}^{-1}$ to its age using the thin disk vertical heating study of Ting & Rix (2019), the probability that this star is from the thin disk is negligible. This demonstrates that the star is most likely from the thick disk and therefore likely more than 8 Gyr old.

We perform a power-spectrum analysis of the TESS light curve presented in Section 2.18 and displayed in Figure 3. We apply the astropy Lomb–Scargle periodogram implementation (VanderPlas 2018) to the light curves from TESS Sectors 11 and 12 considered separately and together. Since strong TESS systematics occur on the timescale of its orbit (13.5 days), these are filtered out at the calibration stage of our analysis and therefore longer-timescale astrophysical variability is also removed.

The resulting power spectra are displayed in Figure 7 and the results are similar for the sectors treated independently. The spectrum decays with a pink noise or "flicker" profile above about 0.1 cycles per day; systematics on this timescale have been filtered out. The "flicker" is consistent with stochastic variability, such as the turbulent motion of a convective photosphere or an accretion disk.

Zoom In Zoom Out Reset image size

Many characteristics of CN Cha seem to suggest we have caught a Young Stellar Object in outburst. In this case, the main YSO candidates would be either a T Tauri star (Appenzeller & Mundt 1989; Miller et al. 2011), a Herbig Ae/Be star (Hillenbrand et al. 1992; Waters & Waelkens 1998), or an FU Orionis type star (Hartmann & Kenyon 1985, 1996). Indeed many new Young Stellar Objects have been discovered using Gaia in combination with archival data (Hillenbrand et al. 2018, 2019). In particular, the timescale and relatively consistent brightness in the light curve post-outburst shown in the top panel of Figure 2 seem to be quite consistent with a period of increased accretion that is commonly seen in FU Ori type stars (Hartmann & Kenyon 1996).

There are two key observations that lead us to believe that a YSO is not a sufficient explanation of the data:

• 1.  
  CN Cha appears to have been a Mira-type long-period variable star from 1963 to as late as 2010, as laid out in Section 4.1. As far as we are aware, there is no YSO object that has photometric variability characteristic of a Mira.
• 2.  
  The dynamics of the star, suggested by Gaia proper motions, indicate that the star is quite old. This is explained above in Section 4.6.

This evidence strongly indicates that CN Cha is not some kind of YSO. Though it is true that young stars appear everywhere in our galaxy (Price-Whelan et al. 2019), they occur with much lower frequency outside of the disk plane (Binney & Tremaine 2008; Belokurov et al. 2020). The fact that, based on Gaia dynamics, the last time that CN Cha crossed the midplane of the disk was 52.5 Myr ago is a particularly bad sign for the YSO theory given our understanding of star formation mostly occurring in dense molecular clouds that are most common in the disk midplane (McKee & Ostriker 2007).

Another possible explanation for the phenomena we have outlined above would be the onset of the protoplanetary nebula (PPN) phase at the end of an evolved AGB cycle (Kwok 1993). While the observations of a massive outflow, combined with CN Cha being luminous in the infrared, and the presence of strong hydrogen recombination lines, are consistent with a planetary nebula, the evolutionary timescales exhibited by CN Cha are simply too short to be associated with a planetary nebula (Marigo et al. 2001, 2004). In particular, PPN theoretically should not show significant dimming from their peak magnitudes until several thousand years after the onset of the PPN phase (Marigo et al. 2001).

Additionally, our spectra of CN Cha exhibit no signs of the [O iii]5007 Å line that is ubiquitously present in all known planetary nebulae (Méndez 2017). This would tend to indicate that our system is much cooler than the typical protoplanetary nebula.

A symbiotic binary star is a binary star system consisting of an evolved star, usually a red giant branch (RGB) star or asymptotic giant branch (AGB) star, and a hot, compact companion, usually a white dwarf (Podsiadlowski & Mohamed 2007; Mohamed & Podsiadlowski 2012). Those symbiotic binaries with RGB primaries are known as S-type systems while those with AGB companions, which are typically Mira variables and are surrounded by a dusty envelope, are known as D-type systems. A systematic, all-sky search for these objects was recently carried out by Akras et al. (2019), characterizing the relative proportion of these types. Binary systems with one component consisting of a white dwarf are thought to be the progenitors of classical novae, wherein hydrogen-rich material that is accreted from the binary companion onto the white dwarf triggers a thermonuclear runaway (TNR, Schatzman 1949, 1951; Gurevitch & Lebedinsky 1957; Cameron 1959; Starrfield et al. 1972, 2016). Though most classical novae are thought to have late-type main-sequence star companions (Darnley & Henze 2020), there are a few nova systems known to contain evolved-star companions, and they particularly seem to make up the majority of known recurrent novae (RN), which have repeating outbursts on human-measurable timescales (Patat et al. 2011; Darnley et al. 2012).

There are many characteristics that make the observations of CN Cha consistent with a nova occurring in a symbiotic binary system. The star clearly was classified as a Mira-type variable (an evolved AGB star; Catelan & Smith 2015) for decades, explained in Section 4.1. The symbiotic binary theory would also explain the UV excess exhibited by the UVOT observations, visible in the bottom left panel of Figure 2, and fit by a 10,000 K blackbody. The P-Cygni-like absorption profiles exhibited in Section 4.4 would also be well explained by the dusty envelopes that are often observed in D-type symbiotic binaries. This would be consistent with the dynamics of the star outlined in Figure 8, which indicate that it is an old thick-disk star. The stochastic variability observed in the TESS light curve and analyzed in detail in Section 4.7 is also well explained by an accretion disk around the white-dwarf companion.

Zoom In Zoom Out Reset image size

The plateau optical luminosity for the light curve of CN Cha as presented by the ASAS–SN data is also typical for symbiotic novae, especially for the class of symbiotic novae known as "slow novae" of which PU Vul is one prototypical examples along with a handful of others Mikolajewska (2010). On the other hand, CN Cha has quite a short plateau timescale compared with other members of this class. We discuss this further below.