eRASSt J074426.3 + 291606: prompt accretion disc formation in a ‘faint and slow’ tidal disruption event Open Access

ABSTRACT

1 INTRODUCTION

A star that passes within the tidal radius of a black hole (BH) will be torn apart by strong gravitational forces, in a so-called tidal disruption event (TDE; e.g. Hills 1975; Rees 1988). Ensuing this, the remaining bound portion of the stellar debris is elongated into long, thin streams, which evolve on highly eccentric orbits around the BH. Whilst aspects of accretion flow formation in TDEs are still uncertain (see Bonnerot & Stone 2020 for a recent review), the disc formation process and subsequent accretion of the stellar debris onto the BH generates luminous electromagnetic flares that evolve over time-scales of weeks-to-years, depending on the configuration of the TDE itself. These events can cause the accretion rate onto the BH to transition from sub-Eddington, to super-Eddington, and back, over time-scales of years, and thus may present excellent tools to probe both accretion physics, as well as the demographics of quiescent BHs.

The first strong TDE candidates (Bade, Komossa & Dahlem 1996; Grupe, Thomas & Leighly 1999; Komossa & Bade 1999; Komossa & Greiner 1999; Greiner et al. 2000) were discovered by ROSAT (Trümper 1982) and were broadly consistent with the observational signatures of TDEs predicted in Rees (1988). Later discoveries of X-ray bright, non-jetted TDEs (Maksym, Ulmer & Eracleous 2010; Lin et al. 2011; Saxton et al. 2012; Maksym et al. 2013; Donato et al. 2014; Khabibullin & Sazonov 2014; Lin et al. 2015; Lin et al. 2017; Saxton et al. 2017), pre-dominantly with XMM-Newton (Jansen et al. 2001) and Chandra (Weisskopf et al. 2000), built up a broad picture of these events as large amplitude, ultra-soft X-ray flares from galaxies that had shown no strong signatures of prior active galactic nucleus (AGN) activity.

Several ultraviolet (UV)-bright TDEs were also identified in the 2000s (Gezari et al. 2006, 2008, 2009) with the Galaxy Evolution Explorer (Martin et al. 2005). Recent years have seen a significant increase in the reported detection rate in TDE candidates, with these having been primarily discovered through optical surveys such as the All Sky Automated Survey for SuperNovae (ASAS-SN; Holoien et al. 2014; Leloudas et al. 2019; Wevers et al. 2019, 2021; Hinkle et al. 2020a), Asteroid Terrestrial-impact Last Alert System (ATLAS; Wang et al. 2022), Panoramic Survey Telescope and Rapid Response System (PanSTARRS; Holoien et al. 2019), Palomar Transient Factory and intermediate Palomar Transient Factory (PTF and iPTF; Arcavi et al. 2014; Blagorodnova et al. 2017), Optical Gravitational Lensing Experiment (OGLE; Wyrzykowski et al. 2017), and Zwicky Transient Facility (ZTF; van Velzen et al. 2019, 2021; Nicholl et al. 2020). Curiously, the vast majority of optically selected TDEs do not show any transient X-ray emission (∼70 per cent of optically selected TDEs in van Velzen et al. 2021 were not X-ray ‘loud’). Possible mechanisms for explaining the dearth of X-rays from optically selected TDEs include the optical emission being powered by stream–stream collisions, instead of accretion (Piran et al. 2015; Shiokawa et al. 2015), or that the high-energy disc emission is being reprocessed to lower wavelengths, either through a quasi-spherical gaseous envelope (e.g. Loeb & Ulmer 1997), or outflows (e.g. disc winds, Lodato & Rossi 2011; Miller 2015; Metzger & Stone 2016; Dai et al. 2018; collision-induced outflows, Lu & Bonnerot 2020). Dai et al. (2018), suggesting that the observed properties of a TDE may be highly dependent on the observer’s viewing angle to the system, with X-ray bright TDEs being observed closer to pole-on inclinations, and optically bright TDEs being observed edge-on.

Launched in 2019, eROSITA (Predehl et al. 2021), on-board the Russian-German SRG mission (Sunyaev et al. 2021), has rapidly expanded the sample of X-ray selected TDE candidates (e.g. Brightman et al. 2021; Malyali et al. 2021; Sazonov et al. 2021, Malyali et al. in preparation). In this paper, we report on a set of multiwavelength observations of the TDE candidate eRASSt J074426.5 + 29160 (herein J0744), associated to a previously inactive galaxy at z = 0.0396. In Section 2, we describe the initial discovery of J0744, before presenting an overview of X-ray observations of J0744 in Section 3, and its photometric evolution in Section 4. We then analyse the host galaxy properties of this TDE in Section 5 and discuss the unique aspects of this event and implications for TDE science in Section 6. We conclude with a summary in Section 7.

We adopt a flat ΛCDM cosmology throughout this paper, with |$H_0=67.7\, \mathrm{km}\, \mathrm{s}^{-1}\mathrm{Mpc}^{-1}$|⁠, Ω_m = 0.309 (Planck Collaboration 2016); z = 0.0396 therefore corresponds to a luminosity distance of 180.5 Mpc. All magnitudes will be reported in the AB system and corrected for Galactic foreground reddening using A_V = 0.0902 mag, obtained from (Schlafly & Finkbeiner 2011) and R_V = 3.1, and using the effective wavelength for each filter retrieved from the SVO Filter Profile Service,¹ unless otherwise stated. All dates/times will be reported in universal time or MJD.

2 DISCOVERY

J0744 was first identified during a systematic search for TDEs in the second eROSITA All-Sky Survey (eRASS2), where it was detected by eROSITA on 2020 October 21 as a new, bright (0.3–2 keV observed flux of ∼1 × 10⁻¹² erg cm⁻² s⁻¹), ultra-soft (photon index ∼4) X-ray point source, which had not been detected 6 months previously in eRASS1 (Section 3.2). Using the eROSITA Science Analysis Software pipeline (eSASS; Brunner et al. 2022), the source was localized to (RA_J2000, Dec._J2000) = (07^h44^m26.3^s, + 29°16^′04.8^″), with a 1σ positional uncertainty of 1.7 arcsec (68 per cent confidence), and thus consistent with a pair of galaxies at z = 0.0396 (Fig. 1). Based only on this localization provided by eROSITA, the optical counterpart was initially unclear. However, on 2020 October 2, the ALeRCE alert broker (Förster et al. 2021) discovered the optical transient, AT 2020uwq/ ZTF20acgrymn², within the Zwicky Transient Facility survey data (ZTF; Bellm et al. 2019; Graham et al. 2019), which is associated with the centre of the smaller galaxy in Fig. 1 (herein SDSS J074426.12 + 291607.4- Section 5). As it is highly unlikely that there also exists an ultra-soft X-ray transient (Section 3) evolving contemporaneously in the more massive galaxy seen in Fig. 1, we consider the transient X-ray emission initially detected by eROSITA to also originate from SDSS J074426.12 + 291607.4 throughout this work.

Figure 1.

Legacy DR8 g, r, z false colour image centred on the host galaxy of J0744. The white cross-hairs and cyan circle mark the ZTF position and eROSITA localization respectively, where the radius of the circle is set to the 2σ uncertainty on the eROSITA source position. The separation between the two galaxy centres is ∼5.4 arcsec, corresponding to a projected distance of ∼4.3 kpc (Section 5.2).

Open in new tabDownload slide

3 X-RAY OBSERVATIONS

3.1 Archival

The position of J0744 was serendipitously observed by Chandra during a 10.8 ks exposure on 2011 February 7 with ACIS-S. No X-ray point source was detected at the position of the host galaxy in this observation. Using the ciao (Fruscione et al. 2006) tool srcflux, the 3σ upper limit on the 0.3–2 keV flux of this source is 2.6 × 10⁻¹⁵ erg cm⁻² s⁻¹, assuming the same spectral model as the best-fitting spectral model to eROSITA spectra described in Section 3.5. The nucleus of the more massive galaxy was detected as a faint point source in the Chandra Source Catalog 2 (CSC2) located at (RA_J2000, Dec._J2000) = (07^h44^m26.5^s, + 29°16^′10.2^″), with a positional uncertainty of 1.3 arcsec (95 per cent confidence) and with 0.5–7 keV flux of (1.57 ± 0.07) × 10⁻¹⁴ erg cm⁻² s⁻¹. However, the X-ray emission from this source is hard, with no source detection reported in the 0.2–0.5 or 0.5–1.2 keV bands in CSC2. Assuming an AGN-like spectral model with Γ = 2 (e.g. Nandra & Pounds 1994) and absorption set to the Galactic neutral hydrogen equivalent column density along the line of sight, the 3σ UL on the source’s 0.3–2 keV flux is 6 × 10⁻¹⁵ erg cm⁻² s⁻¹. In addition to the Chandra observations, no X-ray source was detected within 60 arcsec of J0744 during ROSAT observations during the 1990s, nor during Neil Gehrels Swift X-ray Telescope (XRT; Gehrels et al. 2004; Burrows et al. 2005) observations in 2014. Both of these provide weaker constraints on the archival X-ray flux than the 2011 Chandra observation.

3.2 eROSITA

J0744 was detected by the eROSITA source detection pipeline as a new ultra-soft X-ray source on 2020 October 21 during eRASS2, with a 0.3–2 keV count rate of |$1.05^{+0.14}_{-0.13}$| ct s⁻¹. No source was detected by eROSITA when it scanned the position of J0744 on 2020 April 18 in eRASS1, with a 3σ upper limit on the 0.3–2 keV count rate 0.04 ct s⁻¹. J0744 was also later observed by eROSITA during eRASS3 and eRASS4 between 2021 April 19 and 20, and 2021 October 21 and 22, respectively (Tables 1 and 2). For each observation of J0744 within each eRASS, spectra and light curves were extracted using the SRCTOOL task provided by eSASS (version eSASS_users211214). To extract source spectra, a circular aperture of radius 60 arcsec,³ centred on the eRASS2 position, was used, whilst background spectra and light curves were extracted from a source-free annulus with inner and outer radii of 140 and 240 arcsec, respectively. No significant variability is seen in the 0.2–2.3 keV band within the four observations of J0744 during eRASS2 (Fig. 2).

Figure 2.

0.2–2.3keV band eROSITA light curves of J0744, with each sub plot showing the X-ray evolution within a given eRASS. Within an eRASS scan, each successive visit is ∼4 h after the previous visit, whilst each successive eRASS is ∼6 months after the previous eRASS. The black markers represent the count rate measured within the source aperture (not-background subtracted), whilst grey markers denote the background count rate measured within the background aperture (re-scaled to the size of the source aperture). t − t_{0, i} represents the time taken since the start of observations of J0744 within eRASS i.

Open in new tabDownload slide

The eROSITA 0.3–2 keV count rates were converted to 0.3–2 keV observed fluxes using an energy conversion factor (ECF) of 9.4 × 10⁻¹³ erg cm⁻² and 1.2 × 10⁻¹² erg cm⁻² for the fluxes corrected for Galactic absorption, which were computed using the best-fitting spectral model to the eRASS2 observation (Section 3.5). The 0.3–2 keV flux of this source in the eRASS2 observation is |$9.8^{+1.3}_{-1.2} \times 10^{-13}$| erg cm⁻² s⁻¹ (Table 3), thus the source had brightened by a factor of at least ∼160 relative to the 3σ upper limit from the archival Chandra observation in 2011.

3.3 Swift XRT

Following the eROSITA detection, J0744 was further monitored by the XRT and the Ultra-violet Optical Telescope (UVOT; Roming et al. 2005) instruments on board the Neil Gehrels Swift observatory (Table 1). Observations taken with XRT were performed in photon-counting mode. The XRT light curves were generated and downloaded from the UK Swift Science Data Centre website (Evans et al. 2007, 2009) and were analysed using version 6.29 of the heasoft analysis software, with caldb v20210915. The source count rates and upper limits (Table 4) provided by this tool were inferred using the Kraft, Burrows & Nousek (1991) method, and have already been corrected for pile-up, CCD dead zones, and the loss of photons outside of the source extraction regions (Evans et al. 2009). For the two XRT observations where more than 20 counts were detected within the source extraction region, xrtproducts was also used to generate source and background spectra. To do this, source counts were extracted from a circular aperture of 47 arcsec radius, with background counts extracted from an annulus with inner and outer radii 70 and 250 arcsec, respectively. The 0.3–1 keV XRT count rates were then converted to observed 0.3–2 keV fluxes using an ECF of 2.1 × 10⁻¹¹ erg cm⁻² and 2.1 × 10⁻¹¹ erg cm⁻² for the 0.3–2 keV fluxes corrected for Galactic absorption, which was again computed using the best-fitting spectral model for the XRT observation with OBSID 00013855009 (Section 3.5).

3.4 NICER XTI

J0744 was also monitored over a 4-d window between 2020 November 28 and 2020 December 2 (Table 1) by the X-ray Timing Instrument (XTI) on-board the Neutron Star Interior Composition Explorer (NICER; Gendreau et al. 2016). The observational data were downloaded from the NICER data archive available through heasarc, with which we then ran the nicerl2 task on to generate a set of cleaned event files. For each OBSID, the nibackgen3C50 tool (Remillard et al. 2022) was then used to generate source and (empirically generated) background spectra from the cleaned event files. From these, we inferred total and background count rates in the 0.3–1 keV band, with this band chosen to reduce background contamination of the source count rate due to incomplete background modelling and because of the source’s ultra-soft X-ray spectrum (Section 3.5). J0744 is background-dominated in these observations (Table 5), with 3σ upper limits on the count rates conservatively inferred via CR_tot + 3σ, where σ is the error on the total count rate. The 0.3–1 keV band count rates were then converted to 0.3–2 keV fluxes (Table 3) assuming the best-fitting spectral model to the eRASS2 observation.

3.5 Spectral fitting

The extracted X-ray spectra were analysed using the Bayesian X-ray Analysis (bxa) software (Buchner et al. 2014), which connects the fitting environment xspec (Arnaud 1996) with the nested sampling algorithm ultranest⁴ (Buchner 2021); spectra were fitted unbinned using the C-statistic (Cash 1976).

The eROSITA and XRT backgrounds were first modelled empirically using the principal component analysis technique described in Simmonds et al. (2018; see also Liu et al. 2022a). The extracted source spectra, which contain a contribution from both the source and background emission, were then jointly fitted using the source and background models. However, due to the combination of the relatively faint X-ray flux of the source and the short eROSITA and XRT exposures, only the eRASS2 and Swift observations 00013855008 and 00013855009 had more than 20 counts within the source extraction regions (Tables 2 and 4) and were fitted with bxa.

The eRASS2 X-ray spectrum of J0744 is ultra-soft, with nearly all counts in the source aperture being detected below 2 keV. Five different models were fitted to this spectrum over the 0.2–8 keV range: (i) a redshifted blackbody (tbabs*zbbody), (ii) a redshifted power-law (tbabs*zpwrlaw), (iii) an absorbed redshifted blackbody (tbabs*zabsbb), (iv) an absorbed redshifted power-law (tbabs*zabspwrlaw), and (v) redshifted thermal bremsstrahlung⁵ (tbabs*zbremss). For each fitting, the equivalent Galactic neutral hydrogen column density, N_H, for the tbabs component, was set to its value in the HI4PI survey of 3.52 × 10²⁰ cm⁻² (HI4PI Collaboration 2016), and the redshift was fixed to that of the host galaxy. The same procedure was adopted for the fitting of the XRT spectra, except that these were fitted over the 0.3–8 keV range. The relative values of the Bayesian evidence, computed by UltraNest, is used to choose the best-fitting model to each spectrum (see Buchner et al. 2014), with these listed along with other parameter estimates obtained from bxa in Table 6.

The best-fitting spectral model to the eRASS2 spectrum is tbabs*zpwrlaw, with the inferred photon index, |$\Gamma =4.0 ^{+0.2}_{-0.3}$|⁠, and the convolved source model is plotted in Fig. 3. tbabs*zpwrlaw is also the preferred model for the XRT observation obtained ∼70 d after the eRASS2 observation (OBSID 00013855008), with the photon index, |$3.8 ^{+0.4}_{-0.4}$|⁠, consistent with the eRASS2 observation. Whilst the preferred model for the 00013855009 spectrum is tbabs*zbbody, with |$kT=0.14^{+0.02}_{-0.01}$| keV, overall there is no evidence for any major X-ray spectral changes between the eRASS2, 00013855008 and 00013855009 spectra, when looking at the parameter estimates for a given source model (e.g. kT are consistent with each other for tbabs*zbbody across the three spectra in Table 6).

Figure 3.

bxa fit of the tbabs*zpwrlaw model to the eRASS2 source spectrum of J0744, where the darker (lighter) red filled bands enclose 68 per cent (98 per cent) of the posterior. The X-ray spectrum is ultra-soft, with |$\Gamma =4.0^{+0.2}_{-0.3}$|⁠. Black markers denote the source count rates, with counts being binned to 10 counts per bin for plotting purposes only; the bxa fit was performed on the unbinned spectrum. The solid grey line represents the median of the best-fitting background model inferred from the BXA fit.

Open in new tabDownload slide

3.6 Long-term X-ray variability

Both the XRT and XTI observations improve the sampling of the X-ray light curve, revealing extreme, large amplitude X-ray variability in between the eROSITA observations of J0744 (Fig. 4). To further constrain the evolution of the soft X-ray flux during the XRT monitoring campaign, we stacked the four XRT observations taken between 2020 November 11 and December 3 (XRT observation IDs 00013855001, 00013855003, 00013855004, and 00013855005), and 2021 January 18 and April 10 (XRT observation IDs 00013855011, 00013855012, 00013855013, 00013855014, and 00013855015), summing to net exposures of 6.1 and 4.8 ks, respectively. We assign these two stacked observations the IDs 00013855101 and 00013855102. Fitting the spectra extracted from these two observations with a tbabs*zbbody model with kT fixed to 0.14 keV, then the inferred 0.3–2 keV X-ray flux in these stacked observations are |$2.2^{+1.6}_{-1.1} \times 10^{-14}$| erg cm⁻² s⁻¹ and |$2.6^{+1.9}_{-1.2} \times 10^{-14}$| erg cm⁻² s⁻¹, respectively.

Figure 4.

Multiwavelength evolution of J0744. The top panel shows the X-ray light-curve evolution in the 0.3–2 keV band, with eROSITA (red), NICER (orange), and Swift XRT (blue) data points. The grey horizontal dashed line in this panel marks the 3σ UL on the 0.3–2 keV source flux inferred from a 2011 Chandra pointing, when the source was previously non-detected; the eRASS2 detection represents a brightening by a factor of ∼160 relative to this limit. 3σ upper limits on the source flux during eROSITA and XRT observations are plotted using triangles. The blue crosses in the top panel represent the inferred fluxes from the stacked XRT observations (ObsIDs 00013855101 and 00013855102). The bottom panel shows the ZTF and Swift UVOT light curve. Vertical grey dotted lines denote the times of the optical spectra being taken (Section 5.1).

Open in new tabDownload slide

This implies that the system fades and rebrightens in the 0.3–2 keV band by a factor ∼50 over an 80-d period (Fig. 4). Over the three XRT observations preceding the brightest epoch observed by XRT on MJD 59220, the system brightens by a factor of ∼10 over an 18-d period. This peak brightness is subsequently followed by another drop-off in the X-ray flux and a later rebrightening episode (Fig. 4). At late times (1 yr after the eRASS2 observation), the system is non-detected with eROSITA during eRASS4, and also with XRT.

4 PHOTOMETRIC EVOLUTION

4.1 Optical

We downloaded the publicly available ZTF light-curve data of J0744 using the ZTF forced photometry service (Masci et al. 2019), with the ZTF light curves provided through this interface having been already reference image subtracted. The raw forced photometry light curves were then calibrated using the method developed by Miller et al. (in preparation) for the ZTF Bright Transient Survey,⁶ which involved flagging and removing unreliable measurements, subtracting the pre-outburst baseline fluxes, and re-scaling of flux uncertainties. The ZTF photometry is presented in Fig. 4 and Table A1.

J0744 was observed at peak brightness in the g-band of 19.6 ± 0.2 mag on 2020 October 4, ∼17 d before the eROSITA eRASS2 observation. The optical emission then declined in flux by ∼0.5 mag in the following ∼20 d. The peak observed brightness is faint with respect to both its apparent and absolute magnitude, with the former likely being partially explainable for why the transient was missed in the larger ZTF TDE sample (i.e. it did not show a large amplitude optical flare similar to other TDEs). In terms of its peak absolute g-band magnitude (∼−16.8 mag), it is also the optically faintest TDE reported to date (Fig. 5), out of the TDEs which have had a detection of their transient optical emission. Furthermore, while the other detected TDEs to date with similar peak luminosities have been characterized as ‘faint and fast’ (e.g. iPTF-16fnl, Blagorodnova et al. 2017; AT 2019qiz, Nicholl et al. 2020), the optical light curve of J0744 evolves over much slower time-scales in comparison (Fig. 6). Fitting the r-band light curve with a half-Gaussian function for the rise, and an exponential function for the decay (i.e. following the approach described in section 3.1 of Malyali et al. 2021), then the inferred rise time-scale for the r-band light curve is |$10 ^{+6}_{-3}$| d, and the exponential decay time-scale, τ, is |$120 ^{+15}_{-12}$| d (Fig. 7), thus making J0744 a ‘faint and slow’ TDE (Section 6).

Figure 5.

Comparison of the ZTF g-band light curve of J0744 with other optically bright TDE candidates (blue markers) reported in the ZTF survey (van Velzen et al. 2021; Hammerstein et al. 2022), and also with the faint and fast TDE iPTF-16fnl (Blagorodnova et al. 2017). The peak observed g-band brightness of J0744 is the faintest of all known TDEs identified to date. The line connecting each photometric data point for each TDE is only included to help guide the eye.

Open in new tabDownload slide

Figure 6.

Peak inferred absolute g-band magnitude, M_g, against the exponential decay time-scale, τ, for the population of optically bright TDEs (blue circle markers) presented in tables 1 and 2 of van Velzen et al. (2020). J0744 (red diamond) is extremely faint (M_g ∼ −16.8) relative to the bulk of this TDE population. The ‘faint and fast’ TDEs iPTF-16fnl and AT 2019qiz are highlighted here with a red outline.

Open in new tabDownload slide

Figure 7.

Fit to the calibrated ZTF r-band forced photometry. The inferred rise time-scale for the r-band light curve is ∼10 d, while the exponential decay time-scale, τ, is ∼120 d.

Open in new tabDownload slide

4.2 Swift UVOT

Follow-up observations with the UVOT instrument on-board Swift commenced on MJD ∼ 59164, ∼20 d after the eRASS2 detection. A total of 15 observations were then performed over the following 400-d period, using a mix of the U, UVW1, UVM2, and UVW2 filters. The level 2 UVOT sky images were again downloaded from the UK Swift Science Data Centre, with aperture photometry being performed on these using the uvotsource task (heasoft v6.29, caldb v20201215). Given the close proximity of the more massive galaxy to the host of J0744, a 3 arcsec radius source aperture was used, while background counts were extracted from a nearby source-free region of radius 15 arcsec. The 3 arcsec radius was chosen in order to minimize the contamination from the more massive galaxy. Upon visual inspection, it was noticed that a number of image frames were affected by point spread function (PSF) smearing, which was leading to clearly erroneous photometric data points; these frames were first identified and discarded before the running of uvotsource. Lastly, an aperture correction was then performed on the coincidence-loss corrected count rates using the CURVEOFGROWTH option within uvotsource, and we also performed the recommended Small Scale Sensitivity check before computing the UVOT photometry.⁷ In addition to these follow-up observations, the host of J0744 had previously been serendipitously observed by Swift UVOT in 2014. We first generated a stack of the three Swift observations performed on 2014 April 29 and November 8 and 9, using uvotimsum, and then computed aperture photometry using the same procedure as above (Table 7).

4.3 Photometric light-curve analysis

For each unique MJD that a Swift observation was performed,⁸ we generated synthetic fluxes and errors in the g and r bands using the exponential decay models described in Section 4.1 (if no ZTF photometry was available for that MJD). Then, the r, g, UVW1, UVM2, and UVW2 photometry for each MJD was fitted with a blackbody, assuming a log uniform prior on the blackbody temperature, T_BB, between 10³ and 10⁶ K, and on the radius, R_BB, between 10¹² and 10¹⁷ cm. The evolution of R_BB, T_BB, and the blackbody luminosity, L_BB, inferred via the Stefan–Boltzmann law, is plotted in Fig. 8. No additional extinction in the host galaxy of J0744 is included here.⁹ As the luminosity decreases over time, the temperature remains approximately constant with T_BB = (2.1 ± 0.2) × 10⁴ K, while the radius decreases over time. We also fitted the decay phase of the UV photometry with a power law of the form f_ν = [(t − t₀)/τ]ⁿ (Fig. 9). The inferred n are |$-0.9^{+0.2}_{-0.2}$|⁠, |$-0.83^{+0.17}_{-0.14}$|⁠, and |$-1.02^{+0.13}_{-0.07}$| for the UVW1, UVM2, and UVW2 bands, respectively.

Figure 8.

Evolution of blackbody emission inferred from fitting the optical and UV photometry for J0744.

Open in new tabDownload slide

Figure 9.

Power-law fits to the Swift UV photometry of J0744, with best-fitting slopes |$-0.9^{+0.2}_{-0.2}$|⁠, |$-0.83^{+0.17}_{-0.14}$|⁠, and |$-1.02^{+0.13}_{-0.07}$| for the UVW1, UVM2, and UVW2 bands, respectively. Circular and triangle markers represent measured fluxes and 3σ ULs, respectively, while the shaded bands enclose the inner 68 per cent of the credible region for the fitted model.

Open in new tabDownload slide

4.3.1 MOSFiT

The free parameters of this model are M_bh, M_⋆, b (the scaled impact parameter for the TDE), ϵ, T_viscous, t₀ (the time of first debris fallback measured relative to peak luminosity), R_{ph, 0}, l_ph, and n_{H, host} (the neutral hydrogen column density in the TDEs host galaxy). We adopt the same priors on each parameter as those reported in Mockler et al. (2019), with the exception of the prior on the t₀, which we set to a uniform prior within ±50 d of peak optical luminosity, and M_⋆, for which we use a log-uniform prior between 0.01 and 100 |$\, \mathrm{M}_{\odot }$|⁠. Fits were run using the nested sampler dynesty (Speagle 2020) until convergence, and we performed a simple check of our parameter estimates through ensuring that two successive, differently seeded MOSFiT runs returned consistent values. The fitted MOSFiT TDE model (Fig. 10) reproduces both the early and late-time observed fluxes in the UVW1, UVM2, and UVW2 filters, but the fit to the ZTF r-band data underestimates the flux from ∼80 d after the optical peak. Similar excesses between model and data have been seen in the MOSFiT modelling of other TDEs in the literature (Mockler et al. 2019; Nicholl et al. 2022), suggesting that single-component models for the emission are likely not capable of capturing the full optical-UV evolution of TDEs (potentially due to an additional contribution from the disc emission).

Figure 10.

MOSFiT TDE model fits to the ZTF and Swift UVOT photometry of J0744. The markers and filled in regions are the same as for Fig. 9.

Open in new tabDownload slide

A full list of parameter estimates generated from the posteriors is presented in Table 8. The inferred BH mass is |$\log (M_{\mathrm{bh}}/ \mathrm{M}_{\odot })=6.01 ^{+0.29}_{-0.35}$|⁠, while the stellar mass is |$M_{\star }/ \mathrm{M}_{\odot }=(0.09 ^{+0.03}_{-0.02})\pm$|0.66 (i.e. unconstrained and dominated by systematic error here). The scaled impact parameter is constrained to be |$0.50 ^{+0.25}_{-0.16}$|⁠, suggesting that the TDE was produced by a partial, instead of full, disruption. The peak bolometric luminosity, produced by the reprocessing component, is ∼2 × 10⁴³ erg s⁻¹, and the integrated emitted energy from this is ∼2 × 10⁵⁰ erg, corresponding to a total accreted mass of |$M_{\mathrm{acc}}\sim 0.002 \, \mathrm{M}_{\odot } \left(\epsilon / 0.05 \right)$| (Table 8). It is important to consider that these estimates were inferred from MOSFiT fitting a single temperature blackbody model to the UV photometry and do not model the contribution from the accretion disc in the far-UV to soft X-rays. While such an approach may work for some optically selected TDEs that have their early-time optical emission dominated by stream–stream collisions (e.g. Piran et al. 2015), it is not currently clear how MOSFiT’s parameter estimates are systematically biased through not directly modelling the contribution from the disc to the SED.

5 HOST GALAXY PROPERTIES

5.1 Optical spectroscopy

5.1.1 Archival

The host galaxy of J0744 was observed on 2015 November 10 by the SDSS Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) survey (Bundy et al. 2014), which provides integral field spectroscopy for galaxies with z < 0.15. The data cube for this observation (plate-IFU: 8146-3702) was downloaded from the Marvin web application (Cherinka et al. 2019) and then analysed with the MaNGA Data Reduction Pipeline (DRP; Westfall et al. 2019), using the latest observation summary file (DRPALL v3_1_1) produced by the DRP. We extracted a spectrum from a 0.5 × 0.5 arcsec² spaxel with spaxel coordinates (32,17), covering the Gaia position of J0744, with this showing a quiescent galaxy-like spectrum (Figs 11 and 12). Combined with the non-detection in the archival Chandra observation (Section 3.1), it is then highly likely that J0744 did not harbour a luminous actively accreting supermassive BH prior to the 2020 outburst.

Figure 11.

Optical spectra of the host galaxy of J0744, suggesting that the host was quiescent prior to the optical, UV and X-ray outburst observed in late 2020 November.

Open in new tabDownload slide

Figure 12.

pPXF model fit (red) to the pre-outburst MaNGA spectrum (black) of J0744’s host galaxy.

Open in new tabDownload slide

5.1.2 Follow-up observations

Several spectroscopic observations of J0744 were performed (Table 9), with the first spectrum being obtained with AFOSC mounted at the 1.82-m telescope (OAPd/INAF) of the Asiago observatory ∼52 d after the time of peak observed optical brightness of J0744. It was observed on 2020 November 25 and 26, with VPH7 grism, which covers a range of 3140–7280 Å with a dispersion of about 4.88 Å px⁻¹, and finally on 2020 November 27 with VPH6 grism, which covers a range of 4850–9300 Å with a dispersion of about 5.26 Å px⁻¹. The 1.69 arcsec-slit gave a resolution of about 16.5 Å with VPH7 and about 15.5 Å with VPH6. 1 × 1200 s exposure was obtained on November 25, 2 × 1200 s on November 26, and 4 × 900 s on November 27 under a seeing of about 1.5, 1.7, and 1.3 arcsec, respectively. Spectra were reduced with iraf following the usual procedure of overscan subtraction, flat-field correction, wavelength calibration by means of mercury-cadmium and neon lamps, flux calibration through the observation of the standard star G191-B2B, and finally night-sky subtraction. The three spectra obtained with grism VPH7 were combined after having overlapped them and checked their agreement. Finally, a total 3500–9300 Å spectrum was obtained by adding the two spectral ranges extracted by choosing an aperture of 4 arcsec.

A second observation was then performed ∼4 months later with the ESO Faint Object Spectrograph and Camera (v.2; EFOSC2; Buzzoni et al. 1984) mounted on the ESO New Technology Telescope (NTT) on 2021 March 31 (proposal ID 106.21RU.001, PI: Malyali). We used the grism 13 and the 1.2 arcsec slit oriented 90 degrees off the parallactic angle. We obtained two consecutive exposures, 1800 s each with the seeing around 1 arcsec during the observation. The spectrum has the wavelength range of 3685–9315 A with a dispersion of 2.77 A pixel⁻¹. The data were reduced and calibrated using the esoreflex pipeline (Freudling et al. 2013, v2.11.5). The He + Ar arcs were used to obtain the wavelength calibration, and the standard star LTT3864 was used for flux calibration, which was observed with the same grism and the same slit oriented along the parallactic angle.

5.1.3 Spectral analysis

The pre-outburst MaNGA spectrum was first fitted using the penalized pixel fitting (pPXF) routine (Cappellari & Emsellem 2004; Cappellari 2017), using the stellar template library MILES-HC (Westfall et al. 2019) generated from the MILES library (Falcón-Barroso et al. 2011). The best-fitting model is shown in Fig. 12, from which we infer a stellar velocity dispersion, σ_⋆ = 86 ± 3 km s⁻¹ (corrected for instrumental dispersion following the procedure suggested in Westfall et al. 2019), and log [M_BH/M_⊙] = 6.3 ± 0.1 when using the M_BH–σ_⋆ relation in McConnell & Ma (2013).

Each of the follow-up optical spectra of J0744 show a quiescent galaxy-like spectrum, with no ‘typical’ AGN-like emission lines detected above the host emission (Fig. 11). To search for transient spectral features post-outburst, we downsampled the MaNGA spectrum to the Asiago spectrum, normalized both spectra by the mean of their continuum emission in the 5800–6200 Å range, and then subtracted the MaNGA from the Asiago spectrum (Fig. 13). A transient blue continuum is present in the difference spectrum (Fig. 13), with possible broad emission features being present near 4100 and 4600 Å . Although the difference spectrum is noisy, we attribute the emission feature near 4600 Å to be due to a potential blend of He ii and N iii 4640 Å (highly blueshifted Hβ is disfavoured due to the absence of broad Hα emission; Fig. 11).

Figure 13.

Comparison of the pre-outburst spectrum (MaNGA) with the Asiago spectrum taken ∼52 d after peak optical brightness. Both spectra have been normalized by their flux in the 5800–6200 Å range. The difference between these two spectra is plotted in red, revealing the presence of a transient blue continuum and possible broad emission features around 4100 and 4600 Å .

Open in new tabDownload slide

5.2 Morphology

One of the most striking aspects of J0744 is its host galaxy (Fig. 1), which is part of a pair of quiescent galaxies at z = 0.0396. Using the Gaia EDR3 (Gaia Collaboration 2016, 2021) positions for each galaxy, the separation between this pair is ∼5.4 arcsec, corresponding to a projected distance of ∼4.3 kpc. A detailed analysis of their morphologies has previously been performed using SDSS DR7 imaging in Simard et al. (2011). In this work, the objects were first deblended using the sextractor software (Bertin & Arnouts 1996), before three different models were fitted to the host: (i) a pure Sérsic decomposition, (ii) an exponential disc and de Vaucouleurs bulge (with Sérsic index fixed to 4), (iii) an exponential disc and bulge with Sérsic index left as a free parameter, with the F-statistic then used for selecting the best-fitting model amongst these. The inferred morphological parameters for the host of J0744 and its companion are presented in Table A2, with Simard et al. (2011) finding that an exponential disc and bulge, with n_b = 3.5 ± 0.3, is the best-fitting model for the host of J0744. The inferred half-light radius of the more massive galaxy, ∼4 kpc, is comparable to the projected separation the centres of galaxy pair (∼4.3 kpc).

The Galactic extinction-corrected SDSS g-band model magnitude of the host is 17.89 ± 0.01, translating to an absolute g-band magnitude of −18.39 ± 0.01. Using this, along with the half-light radius of J0744’s host (∼1.8 kpc; Simard et al. 2011), and comparing to the population of dwarf galaxies¹⁰ reported in fig. 6 of Torrealba et al. (2016), one sees that the host of J0744 is at the more luminous, and extended, end of known dwarf galaxies, and has similar luminosity and half-light radius to the Large Magellanic Cloud.

Lastly, we used the relation between M_BH and absolute r-band bulge magnitude, |$M_{\mathrm{R,\, bulge}}$|⁠, presented in equation 5 of (McLure & Dunlop 2002), to infer log (M_BH/M_⊙) = 6.2 ± 0.6 for the disrupting BH in J0744, consistent with M_BH inferred via the M_BH–σ_⋆ relation. In doing this, |$M_{\mathrm{R,\, bulge}}$| was computed using the SDSS r-band model magnitude and the bulge-to-total fraction of 0.51 in Simard et al. (2011).

6 DISCUSSION

Due to the combination of the large amplitude, ultra-soft X-ray flare (Section 3), the optical variability (Section 4), and the optical spectrum that shows no strong signs of past AGN activity in the host galaxy (Section 5.1), and their strong similarity to other TDE candidates reported in the literature, we consider a TDE as the most suitable explanation for the extreme variability seen in this system. Assuming such an interpretation, the eROSITA detection of ultra-soft X-ray emission within ∼20 d of the observed optical peak, which is assumed to be emitted from the innermost regions of the nascent accretion disc, suggests that the disc formed promptly after disruption in this TDE (in contrast to systems that show evidence for delayed disc (in contrast to systems that show evidence for delayed disc formation, e.g. Gezari, Cenko & Arcavi 2017; Malyali et al. 2021; van Velzen et al. 2021; Chen, Dou & Shen 2022).

6.1 Short time-scale, large amplitude X-ray variability in TDEs

The 0.3–2 keV band light curve of J0744 shows a fading, and then rebrightening, by a factor ∼50 during an 80-d period following the eRASS2 detection (Section 3.6, Fig. 4). This extreme variability is only seen in the X-rays, with the optical-UV flux continuing its monotonic decline during this period, likely suggesting a different physical origin for these two components. Other similar cases of large amplitude X-ray variability, over short time-scales (∼d), have now also been reported in a number of non-jetted TDEs in the literature (SDSS J120136.02 + 300305.5, Saxton et al. 2012; OGLE16aaa, Kajava et al. 2020; Shu et al. 2020; AT2019ehz, van Velzen et al. 2021), with their soft-band light curves shown together in Fig. 14. We note that such variability differs from TDEs that show a gradual late-time X-ray brightening relative to the optical peak (e.g. increases by a factor of ∼10 over a ∼200-d period), such as in ASAS-SN 15oi (Gezari et al. 2017) and AT2019ahz (Hinkle et al. 2020a; van Velzen et al. 2021; Goodwin et al. 2022; Liu et al. 2022b). While we do not place any statistical constraints on the prevalence of this short-time-scale X-ray variability in TDEs in this work, we highlight here that similar cases of extreme variability may be relatively common in the early stages of X-ray bright TDEs.

Figure 14.

Compilation of soft X-ray light curves for TDEs reported in the literature that show large amplitude variability over short time-scales, with data taken from Shu et al. (2020) for OGLE16aaa, Saxton et al. (2012) for SDSS J120136/02 + 300305.5, the Swift XRT for AT 2019ehz, with redshift reported in van Velzen et al. (2021). All luminosities are corrected for Galactic absorption and plotted in the 0.3–2 keV band, except for SDSS J120136.02 + 300305.5, which is in the 0.2–2 keV band. Triangle markers denote 3σ upper limits.

Open in new tabDownload slide

We would disfavour this variability to be due to obscuration of the inner disc by the unbound stellar debris generated in the initial disruption, as this has been predicted to subtend small solid angles measured with respect to the BH (e.g. Krolik et al. 2016). For the flaring in OGLE16aaa, an alternate mechanism suggested to explain the variability was that the late-time flare in 0.3–2 keV band was the result of the geometric thinning of a slim disc (Kajava et al. 2020; Wen et al. 2020), where, as the disc thinned, then the amount of self-obscuration of its innermost regions would decrease. Again, such an origin would be disfavoured at least for J0744 and AT2019ehz, since these systems are bright at early times, fade, and then rebrighten, which would disagree with the net brightening predicted under the thinning disc scenario.

Other possible mechanisms for driving such extreme variability fall into three main categories, consisting of the TDE occurring in a supermassive BH binary, inhomogeneous accretion flows, and the reprocessing of the X-ray emission from the disc. The former has previously been suggested in the literature to explain the light curves of OGLE16aaa (Shu et al. 2020) and SDSS J120136.02 + 300305 (Liu, Li & Komossa 2014), where the authors suggest that the presence of the secondary BH in the binary interrupts the accretion flow onto the primary, causing large amplitude dips in the soft X-ray flux.

Alternatively, these light curves may be a byproduct of the extreme nature of the accretion flows formed in TDEs, with the non-steady-state ‘discs’ fluctuating in effective temperature and observed 0.3–2 keV flux. If this is the case, then a timing analysis of the X-ray light curves could be used as a signature to distinguish TDEs from non-TDE-induced accretion flares in future studies. One caveat to such an origin, as also noted in van Velzen et al. (2021), is that for both J0744 and AT 2019ehz, the 0.3–2 keV flux fades, and then rebrightens back to a flux consistent with the previous peak observed flux. It is not immediately clear why such a rebrightening behaviour is observed if the X-ray variability is produced by random fluctuations in the disc temperature (e.g. Mummery & Balbus 2022).

One potential avenue to explain this latter point would be through assuming that the disc’s luminosity initially remains approximately constant over time, such as in an Eddington-limited plateau phase, but the amount of reprocessing of the disc’s X-ray emission along the observer’s line of sight varies over time. The exact nature of such a reprocessor is currently unclear, but could potentially be a disc wind (e.g. Metzger & Stone 2016; Dai et al. 2018), outflows launched from stream–stream collisions (Lu, Beniamini & Bonnerot 2020), or absorption by a gaseous envelope surrounding the TDE (Loeb & Ulmer 1997). If the envelope provided only a partial covering of the BH (e.g. was clumpy), then the rebrightening episodes would thus coincide with the observer momentarily seeing the innermost regions of the accretion disc through the gaps in this envelope.

6.2 ‘Faint and slow’ TDEs

A number of optically selected TDEs in the literature have now been dubbed ‘faint and fast’ TDEs. The prototype of this class, iPTF-16fnl (Blagorodnova et al. 2017), was initially identified as an outlier relative to the observed optically bright TDE population, with respect to both its peak optical luminosity ((1 ± 0.15) × 10⁴³ erg s⁻¹, around an order-of-magnitude fainter than other TDEs), and short rise and decay time-scales in its light curve (exponential decay time scale ∼15 d). Additional members of this class were later suggested through observations of AT 2019qiz (Nicholl et al. 2020) and AT2018ahl/ATLAS18mlw (Hinkle et al. 2022). This set of ‘faint and fast’ TDEs, considered within the context of the broader population of optically selected TDEs, has motivated the suggestion that TDEs with lower peak optical luminosities will show faster rise and decay time-scales in their light curves (e.g. Blagorodnova et al. 2017; Hinkle et al. 2020b). It is currently unclear why these ‘faint and fast’ events seem to evolve differently to other known TDEs, although it is important to note that these host lower mass supermassive black holes (SMBHs, around |$10^{6}\, \mathrm{M}_{\odot }$|⁠), and the mass fallback time-scale scales as |$t_{\mathrm{fb}} \propto M_{\mathrm{BH}}^{1/2}$|⁠. It has also been suggested that these may be due partial TDEs, which may produce mass fallback rates that decline steeper than the ‘canonical’ t^−5/3 rate (Guillochon & Ramirez-Ruiz 2013; Coughlin & Nixon 2019; Ryu et al. 2020).

Similarly to the ‘faint and fast’ TDEs, J0744 shows (i) a short rise time-scale in its optical light curve, (ii) a faint peak optical luminosity, and (iii) is also hosted by a low-mass SMBH. However, its optical light curve decays over much longer time-scales (Section 4), and with a peak observed 0.3–2 keV luminosity of 5 × 10⁴³ erg s⁻¹, J0744 is also ∼2 × 10⁴ and ∼10³ brighter than iPTF-16fnl (∼2 × 10³⁹ erg s⁻¹) and AT 2019qiz (∼5 × 10⁴⁰ erg s⁻¹), respectively. The X-ray spectra of J0744 are also ultra-soft (Section 3.5), contrasting with the hard spectrum of AT 2019qiz (⁠|$\Gamma =1.1^{+0.6}_{-0.4}$|⁠; Nicholl et al. 2020) of uncertain origin, since it is clearly distinct from the X-ray spectra of other thermal TDEs. J0744 thus shows direct evidence for prompt disc formation only ∼20 d after optical peak (again assuming that the ultra-soft X-ray emission originates from the newly formed disc), whereas AT 2019qiz and iPTF-16fnl only show indirect evidence via the transient Bowen emission lines in their optical spectra; the presence of these lines has previously been attributed to the signature of obscured accretion (Leloudas et al. 2019) when observed in the absence of far-UV/ X-rays from the accretion disc.¹¹We note that we classify the optical light curve of J0744 as showing ‘faint and slow’ behaviour. Given the uncertainty associated with the bolometric correction here, it is not clear that this optical faintness necessarily extends to the bolometric luminosity. For instance, Mummery (2021) find that disc-dominated TDEs radiate only a fraction (∼1 per cent) of the disc’s luminosity in the optical/ near-UV and X-ray bands, with the majority instead being released in the FUV (see also Zanazzi & Ogilvie 2020). This may be important to consider in any future works modelling the multiwavelength evolution of J0744-like events.

Following Metzger & Stone (2016), the ‘faint and slow’ nature of J0744 may potentially be caused by photon trapping in an ionized outflow. If the optical depth to electron scattering is sufficiently high, the photon diffusion time-scale may be less than the expansion time-scale for the wind, leading to photons being ‘trapped’ in such an outflow. Photons then advect outwards up to the trapping radius, defined as the radius where the diffusion time-scale equals the expansion time-scale. During the advection phase, the temperature decreases adiabatically. For systems with |$M_{\mathrm{BH}}\lesssim 10^{7}\, \mathrm{M}_{\odot }$|⁠, adiabatic losses may also cause a suppression of the peak optical luminosity and also a slowed optical light-curve decay at early times (Metzger & Stone 2016). If X-ray selected TDEs are systematically observed with viewing angles similar to J0744, where trapping effects are important for evolution of the optical luminosity, X-ray selected/ bright TDEs may show slower decay time-scales on average than populations of X-ray dim TDEs.

Alternatively, given that the most prominent difference between J0744 and other ‘faint and fast’ TDEs is the detection of luminous X-ray emission in J0744, the differences in the optical light-curve evolution may be linked to the circularization efficiency in each TDE. Given the prompt disc formation in J0744, it is likely that there was efficient circularization of the stellar debris, potentially akin to the runaway circularization seen in recent numerical simulations (Steinberg & Stone 2022). The optical rise of J0744 may be powered by the circularization process, while the optical decay may be dominated by accretion, either directly from the disc or reprocessed emission (as discussed above). In these systems, the ultra-soft X-ray emission from the newly formed disc may still be affected by time-variable reprocessing (e.g. neutral absorption, or trapping in ionized outflows), hence may show a highly non-monotonic decline (Section 6.1). These systems may also still appear X-ray dim in the first months to years after disruption depending on the observer’s viewing angle (Dai et al. 2018).

The corollary is that there would be slowed disc formation and inefficient circularization in ‘faint and fast’ TDEs, where the initial debris may instead settle into an elliptical disc during its infancy. The optical transient emission for these events would then be powered pre-dominantly by outflows launched during the circularization process (Nicholl et al. 2020), instead of accretion. The circularization efficiency of the debris may increase over time (e.g. Steinberg & Stone 2022), and a fraction of the ‘faint and fast’ TDEs may also show a delayed brightening in the soft X-rays as a result of the slowed accretion in these systems. Although again sensitive to reprocessing and disc physics (e.g. the disc’s SED and Comptonization; Mummery & Balbus 2021), the late-time X-ray evolution of these systems may appear similar to the decades-long TDE candidate 3XMM J150052.0 + 015452 (J1500) reported in Lin et al. (2017). As there was not deep, high-cadence optical photometry monitoring J1500 in the years before its X-ray flaring, an optical transient associated with that event may easily have been missed. A caveat of this scenario is explaining how the Bowen fluorescence lines are produced in ‘faint and fast’ TDEs if the circularization is inefficient (although beyond the scope of this work, we note that this may be potentially alleviated by X-rays produced by compressional shocks near pericentre e.g. Zanazzi & Ogilvie 2020; Steinberg & Stone 2022).

7 SUMMARY

We have reported on a set of multiwavelength observations of the TDE candidate eRASSt J074426.3 + 291606, located in the nucleus of a quiescent galaxy at z = 0.0396. The main observed features are as follows:

• J0744 was detected by eROSITA in eRASS2, where it had brightened in the 0.3–2 keV band by a factor ≳160 relative to an archival 3σ upper limit inferred from Chandra observations. The peak observed 0.3–2 keV luminosity is ∼5 × 10⁴³ erg s⁻¹, the X-ray spectrum in eRASS2 is ultra-soft (Γ ∼ 4), which remains soft throughout the ∼400 d monitoring campaign.

• eROSITA, NICER XTI, and Swift XRT observations reveal a net decline in the 0.3–2 keV X-ray flux over a ∼400 d monitoring campaign by at least a factor of 10.

• The 0.3–2 keV band flux fades and rebrightens by a factor ∼50 over an 80-day period, deviating significantly from a smooth t^−5/3 decay traditionally expected for TDEs. We considered a set of large amplitude X-ray variability in TDEs and suggest that such variability may not be uncommon in the early stages of X-ray bright TDE light curves. This non-monotonic declining behaviour will be important to consider in future work computing the X-ray luminosity functions of TDEs (e.g. when inferring the peak X-ray luminosity from an infrequently sampled X-ray light curve).

• J0744 is accompanied by an optical-UV flare, which localizes this transient to the nucleus of the quiescent galaxy SDSS J074426.12 + 291607.4. The detection of ultra-soft X-rays only 20 d after peak optical brightness suggests the prompt formation of an accretion disc in this TDE. Although this optical flare was detected by ZTF, J0744 was missed out from the recent sample of ZTF-selected TDEs (van Velzen et al. 2021; Hammerstein et al. 2022).

• The peak optical luminosity of J0744 is extremely faint (peak absolute g-band magnitude ∼−16.8 mag). While the optical g-band emission makes J0744 the faintest TDE observed to date, it is not ‘faint and fast’ like other optically selected TDEs of similar peak luminosity; instead, it is ‘faint and slow’ (Section 6.2).

Higher redshift variants of this system, or for systems where the host galaxy is even fainter relative to the more massive nearby galaxy than in J0744, may appear as off-nuclear TDEs in future surveys. To date, J0744 is the second strong, off-nuclear TDE candidate reported in the literature (3XMM J215022.4-055108; Lin et al. 2018), with this system also being X-ray bright. X-ray and UV surveys may be particularly well suited to identifying intermediate mass black holes in galaxies, benefiting from the lower amount of contamination from other transient classes (such as supernovae), and due to the transient UV emission being relatively long lived and easier to detect over the quiescent host than at optical wavelengths. Similar events to J0744 will likely be detected by future planned missions, such as with the Einstein Probe (Yuan et al. 2018), which will also provide high-cadence monitoring capabilities for events over the 0.3–10 keV range. In the UV, then TDEs similar to J0744 may be exceptionally well studied with the Ultraviolet Transient Astronomy Satellite (Sagiv et al. 2014), scheduled for launch in 2025, and later also with the Ultraviolet Explorer (Kulkarni et al. 2021). These future planned observatories may therefore prove an invaluable tool to uncover TDEs occurring in BHs with masses |$\lesssim 10^{6}\, \mathrm{M}_{\odot }$| and for probing the BH occupation fraction at the low mass end.

ACKNOWLEDGEMENTS

AM is grateful to both the Swift and NICER teams for approving the many ToO requests, and the Swift team for their assistance with reducing the UVOT data, in particular to Peter Brown. AM would also like to thank Matt Nicholl for assistance with using MOSFiT. AM acknowledges support by Deutsches Zentrum für Luft- und Raumfahrt (DLR) under the grant 50 QR 2110 (XMM_NuTra, PI: Z. Liu). We would like to thank the referee for a constructive report that improved the quality of the paper.

This work is based on data from eROSITA, the soft X-ray instrument aboard SRG, a joint Russian-German science mission supported by the Russian Space Agency (Roskosmos), in the interests of the Russian Academy of Sciences represented by its Space Research Institute (IKI), and the Deutsches Zentrum für Luft- und Raumfahrt (DLR). The SRG spacecraft was built by Lavochkin Association (NPOL) and its subcontractors, and is operated by NPOL with support from the Max Planck Institute for Extraterrestrial Physics (MPE).

The development and construction of the eROSITA X-ray instrument was led by MPE, with contributions from the Dr. Karl Remeis Observatory Bamberg & ECAP (FAU Erlangen-Nuernberg), the University of Hamburg Observatory, the Leibniz Institute for Astrophysics Potsdam (AIP), and the Institute for Astronomy and Astrophysics of the University of Tübingen, with the support of DLR and the Max Planck Society. The Argelander Institute for Astronomy of the University of Bonn and the Ludwig Maximilians Universität Munich also participated in the science preparation for eROSITA.

The eROSITA data shown here were processed using the esass software system developed by the German eROSITA consortium.

This work made use of data supplied by the UK Swift Science Data Centre at the University of Leicester.

The ZTF forced-photometry service was funded under the Heising-Simons Foundation grant #12540303 (PI: Graham).

DH acknowledges support from DLR grant FKZ 50 OR 2003. MK is supported by Deutsche Forschungsgemeinschaft (DFG) grant KR 3338/4-1.

DATA AVAILABILITY

The eRASS1-4 data taken within the German half of the eROSITA sky is currently planned to be made public by Q2 2024. The Swift data is available to download through the UK Swift Data Science website,¹² whilst the NICER data is accessible through NASA’s HEASARC interface.¹³ Publicly available ZTF data can be accessed through the ZTF forced photometry service.¹⁴ Follow-up optical spectra will likely remain private at least until the release of the forthcoming eROSITA-selected TDE population paper, but could be made available upon reasonable request.

Footnotes

REFERENCES