AT 2023clx: the Faintest and Closest Optical Tidal Disruption Event Discovered in Nearby Star-forming Galaxy NGC 3799

We initiated optical $ugri$ band follow-up observations of AT 2023clx with the 1.0m Las Cumbres Observatory Global Telescope network (LCOGT; Brown et al. 2013) immediately after the event was reported on the Transient Name Server on 2023-02-26 UT (Taguchi et al. 2023). We used PanSTARRS (Flewelling et al. 2020) $gri$ band stack images as reference images and employed HOTPANTS (Becker, 2015) for image subtraction. Prior to subtraction, we removed cosmic rays and aligned the images using Astrometry.net. After image subtraction, we performed point spread function (PSF) photometry on the difference image using the Photutils package of Astropy (Astropy Collaboration et al., 2022) for the $gri$ data. For the $u$ band, we performed aperture photometry with a 5″ aperture and then subtracted the host galaxy magnitude ($16.69\pm 0.02$ mag) with the same 5″ aperture measured in the rescaled Sloan Digital Sky Survey (SDSS; Gunn et al. 2006) $u$ band image.

We tried to measure the position of AT 2023clx in the difference image of the LCO $r$-band image obtained on 2023-02-26 UT and the centroid of its host galaxy in the PanSTARRS $r$-band reference image by measuring the barycenter using the SExtractor. The offset between them was measured to be $0.21\pm 0.17$ arcsec, taking into account the uncertainty during the alignment of the image. This corresponds to a physical offset of $49\pm 40$ pc at the distance of NGC 3799. Therefore, we can conclude that AT 2023clx is consistent with the center of the galaxy at the resolution level of the LCO images, making it a potential candidate for TDE.

We were intrigued by this event because we believed that it was still in the rising stage, given that its luminosity was 2 magnitudes fainter than the average peak luminosity of optical TDEs (Yao et al., 2023). However, our initial two observations indicated that it was already in decline, suggesting that it might be a rarely-discovered faint TDE. In fact, its peak luminosity in the $g$ band ($M_{g}=-17.16$) is one of the faintest TDEs discovered thus far, comparable to that of iPTF16fnl (Blagorodnova et al. 2017; Brown et al. 2018; $M_{g}=-17.20$).

UV images were obtained with the Neil Gehrels Swift Observatory (hereafter Swift) with the Ultra-Violet Optical Telescope (UVOT). The Swift photometry (PIs: Gomez, Huang, Leloudas, and Wevers) were measured with UVOTSOURCE task in the Heasoft package using 5″ apertures after subtracting the galaxy background using Swift/UVOT images taken on July 18, 2010. It was placed in the AB magnitude system (Oke & Gunn, 1983), adopting the revised zero points and the sensitivity of Breeveld et al. (2011).

The X-ray Telescope (XRT) photometry was performed using XRTPIPELINE and XRTPRODUCTES. We used a circle with a radius of 47.1″ as the source region and an annulus with an inner radius of 100″ and an outer radius of 200″ as the background. The source was X-ray faint in the observations, and we assumed an absorbed power-law spectrum with an index of $\Gamma=1.75$ (Ricci et al., 2017) and a Galactic hydrogen density of $2.51\times 10^{20}\,\text{cm}^{-2}$ (HI4PI Collaboration et al., 2016). We then derived the 3$\sigma$ upper limit for the flux in the 0.3–10.0 keV range using WebPIMMS¹¹1https://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3pimms/w3pimms.pl. In addition, we stacked the 26 event files ranging from MJD 60002.74 to 60089.23, deriving the upper limit of the mean luminosity of $9.53\times 10^{39}$ erg s^-1 in 3$\sigma$ level, with a total exposure time of 35.79 ks.

We also collected host-subtracted light curves of AT 2023clx from public time domain surveys, including data from the Asteroid Terrestrial Impact Last Alert System (ATLAS; Tonry et al. 2018; Smith et al. 2020), the Zwicky Transient Facility (ZTF; Masci et al. 2019) and the All Sky Automated Survey for SuperNovae (ASAS-SN, Shappee et al. 2014; Kochanek et al. 2017).

The ATLAS $c-$ and $o-$band light curves were obtained using the ATLAS Forced Photometry Service, which produces PSF photometry on the difference images. ATLAS has $3-4$ single exposures within each epoch (typically within one day), so we binned the light curve every epoch to improve the signal-to-noise ratio (SNR). The ZTF light curves were obtained using the Lasair alert broker²²2Website link: https://lasair-ztf.lsst.ac.uk/ (Smith et al. 2019). The ASAS-SN host-subtracted g-band light curves were obtained using the ASAS-SN Sky Patrol photometry pipeline. We also collected and binned the photometry data of ASAS-SN. Considering the depth of g band of ASAS-SN is roughly 18.5 mag (de Jaeger et al. 2022), we only use the photometry data brighter than 18 mag because the data fainter than 18 mag have larger error than our LCO and ZTF photometry.

All light curves, after correction for Galactic extinction, are shown in Figure 1. We assumed a Cardelli et al. (1989) extinction law with $R_{V}=3.1$ and a Galactic extinction of $E(B-V)=0.0268\pm 0.0003$ mag (Schlafly & Finkbeiner 2011).

For better comparison of the peak $g$-band luminosity between AT 2023clx and the ZTF TDE samples, we tried to fit the profile of the light curve profile of AT 2023clx following the method of Yao et al. (2023) (see Section 3.2 for details). The rise function was poorly constrained with either Gaussian or power-law function due to only three detected points in ASAS-SN. Hence, we combined the upper limit to estimate the rise function. Shown in Figure 1, AT 2023clx can roughly be described by a rise in power law (index n = 0.64) and a $(t-t_{0})^{-5/3}$ power-law decline ($t_{0}=\rm MJD\,59977\pm 2$ d) with a peak $g$-band luminosity of log$L_{g}=$ 42.31, which is the faintest TDE discovered by the ZTF survey to date compared to Yao et al. (2023).

Five spectra were acquired for AT 2023clx, including one from SDSS DR7, one observed by Taguchi et al. (2023) with the SEIMEI telescope³³3 We collected this spectrum from TNS website: https://www.wis-tns.org/object/2023clx on 2023-02-26 UT, one observed by the ZTF group with the Keck telescope⁴⁴4 ZTF group upload this spectrum to another transient by mistake: https://www.wis-tns.org/object/2018meh on 2023-03-20 UT and two obtained by ourselves with YFOSC onboard the LiJiang 2.4m telescope (Fan et al. 2015; Wang et al. 2019) on 2023-03-03 UT and 2023-03-13 UT. We reduced the LJT spectra according to the standard procedure for long-slit spectra with PyRAF, a Python-based package of IRAF(Tody 1986, 1993). All spectra are shown in Figure LABEL:spec_all.

First, we use the package SUPERBOL (Nicholl 2018) to fit the spectral energy distribution (SED) of AT 2023clx with a blackbody model. The observing cadence after MJD 60040 in $r,\ i$ and $u$ bands is quite sparse. So we use third-order polynomials only to interpolate the $g,\ c$ and $o$ bands light curve in the late phase to match with the Swift epochs. The evolution of blackbody luminosity, temperature, and effective radius are shown in Figure LABEL:bbfit, compared with other faint TDEs.

The optical/UV light curves of TDEs usually show a monthly rise followed by a decay with a timescale of months to years, sometimes following a $t^{-5/3}$ power-law decline (Gezari 2021; van Velzen et al. 2021). In the case of AT 2023clx, it exhibited a similar decline behavior that was consistent with a $t^{-5/3}$ power-law decline in both the $g$ band and the bolometric light curves (see Figures 1 and LABEL:bbfit). The blackbody temperature was approximately 11,000–13,000 K, which slowly decreased at the peak and was followed by a slow return to 13,000 K. The blackbody radius $R_{BB}$ was approximately $4.61\times 10^{14}$ cm at the peak, followed by a decline. It is worth noting that AT2023clx has the lowest temperature among these faint TDEs, albeit with a medium blackbody radius. As a result, the maximum blackbody luminosity of AT 2023clx is only $(4.56\pm 0.53)\times 10^{42}$ erg s^-1, making it even fainter than previous low-luminosity TDEs (e.g., iPTF16fnl; Blagorodnova et al. 2017, AT 2019qiz; Nicholl et al. 2020, AT 2020wey; Charalampopoulos et al. 2023). Furthermore, these faint TDEs (e.g., AT 2020wey; Charalampopoulos et al. 2023) all exhibit fast declining light curves that are much steeper than the canonical $t^{-5/3}$. In the case of AT 2023clx, the decline appears to be less extreme than that of this subclass of TDEs. However, the limited number of samples at present cannot reveal their true statistical characteristics.

We then compared the optical spectra of these low-luminosity TDEs in Figure 4. Clearly, all these spectra near the peak display a blue continuum with broad $\rm H_{\alpha}$ and $\rm H_{\beta}$ emission, as well as the typical TDE emission line He II 4686 Å. The broad $\rm H_{\alpha}$ is blended with He I 6678 Å, and broad He I 5876 Å line was also evident in the +26 d Keck spectrum. The full width at half maximum (FWHM) of the broad Balmer lines is $\approx$15,000 km s^-1. The $\rm H_{\alpha}$ profile is asymmetric, with a blueshifted peak at +4.3 d and a redshifted peak at +26 d (see Figure 4). A sharp peak on the blue side coincides in wavelength with [O I] 6300 Å, which would indicate the presence of low ionization, low density, and slow-moving gas. The red side may include some contribution from He I 6678 Å, but this is likely not a major contributor, as we see only a very weak He I 5876 Å, similar to AT 2019qiz (Nicholl et al. 2020). Nicholl et al. (2020) eventually attributed the evolution of such $\rm H_{\alpha}$ profiles of AT 2019qiz to an outflow. In addition, as we have seen the asymmetric broad $\rm H_{\alpha}$ profile, the N III 4640 Å also could be blended with He II and $\rm H_{\beta}$. Furthermore, most of the spectra in Figure 4 do not cover the range of N III 4100 Å, and it is hard to confirm whether the weak feature at +26 d is real or not. Thus, the N III 4100 Å can not be ruled out either.

However, the lack of low-ionization metal features (e.g., oxygen and calcium) in spectra near the peak almost excludes the typical Type-II supernovae scenario (Filippenko 1997). Type-IIn supernovae also show strong broad Balmer lines, but sometimes with high ionization lines of intermediate width that can be regarded as a result of CSM interaction (e.g., SN 2005ip, SN 2006jd and SN 2010jl; Stritzinger et al. 2012; Fransson et al. 2014). If the CSM interaction of Type-IIn is relatively weak compared to the above three, the Fe II and Ca II P-Cygni lines appear again in their spectra (Taddia et al. 2013). Furthermore, the optical light curves of Type-IIn are relatively long-lived compared to AT 2023clx and normal supernovae do not stay at a temperature as high as 13,000 K for months.

Therefore, AT 2023clx is consistent with a faint TDE scenario in both aspects of light curves and spectroscopic properties. Based on this, we classify AT 2023clx as a robust faint hydrogen and helium (H+He) optical TDE.

NGC 3799, the host of AT 2023clx, is a well-resolved face-on spiral galaxy (see top panels of Figure 1). The pre-outburst SDSS spectra centered on the galaxy center have placed it in the locus of low-ionization nuclear emission-line region region (LINER) in the Baldwin-Phillips-Terlevich (BPT) diagram (Baldwin et al. 1981; Veilleux & Osterbrock 1987) based on its narrow-line ratios, indicating weak active galactic nuclear (AGN) activity albeit without broad emission lines.

Ramos Padilla et al. (2020) performed a spectral energy distribution (SED) analysis of a sample of 189 nearby galaxies, including NGC 3799. Their fitting considered the AGN emission and obtained a stellar mass of log$(M_{*}/M_{\odot})=9.87\pm 0.02$ and a star formation rate (SFR) of $\rm logSFR=-0.09\pm 0.02$ for NGC 3799. Consequently, it is located on the main sequence of star-forming galaxies (SF) (Chang et al. 2015; see Figure LABEL:SF_sequence). AT 2023clx thus belongs to a TDE in a typical SF galaxy, whereas known optical TDEs show a preference for post-starburst (or green valley) galaxies (Arcavi et al. 2014; French, Arcavi & Zabludoff 2016; French et al. 2020; Hammerstein et al. 2021). Additionally, the fractional AGN contribution of NGC 3799 is indeed small ($f_{AGN}\textless 0.2$) according to their fitting. We emphasize that the overall star formation activity does not conflict with its LINER classification in the BPT diagram since the central regions of the SF galaxies are commonly found to be quenched (Tacchella et al. 2015; Ellison et al. 2018). Using the empirical relation between $M_{\mathrm{BH}}$ and the total galaxy stellar mass in the local universe (Reines & Volonteri 2015):

The central $M_{\mathrm{BH}}$ in NGC 3799 is estimated to be $10^{6.26\pm 0.28}M_{\odot}$, using $\alpha=7.45\pm 0.08$ and $\beta=1.05\pm 0.11$.