Infrared Echoes of Optical Tidal Disruption Events: ~1% Dust Covering Factor or Less at sub-parsec Scale
Ning Jiang, Tinggui Wang, Xueyang Hu, Luming Sun, Liming Dou, Lin Xiao
DD RAFT VERSION F EBRUARY
17, 2021Typeset using L A TEX twocolumn style in AASTeX63
Infrared Echoes of Optical Tidal Disruption Events: ∼
1% Dust Covering Factor or Less at sub-parsec Scale N ING J IANG ,
1, 2 T INGGUI W ANG ,
1, 2 X UEYANG H U ,
1, 2 L UMING S UN , L IMING D OU , AND L IN X IAO
1, 21
CAS Key laboratory for Research in Galaxies and Cosmology, Department of Astronomy, University of Science and Technology of China, Hefei, 230026, China;[email protected] School of Astronomy and Space Sciences, University of Science and Technology of China, Hefei, 230026, China; [email protected] Department of Physics, Anhui Normal University, Wuhu, Anhui, 241000, People’s Republic of China Department of Astronomy, Guangzhou University, Guangzhou 510006, China
ABSTRACTThe past decade has experienced an explosive increase of optically-discovered tidal disruption events (TDEs)with the advent of modern time-domain surveys. However, we still lack a comprehensive observational view oftheir infrared (IR) echoes in spite of individual detections. To this end, we have conducted a statistical studyof IR variability of the 23 optical TDEs discovered between 2009 and 2018 utilizing the full public dataset of
Wide-field Infrared Survey Explorer . The detection of variability is performed on the difference images, yieldingout 11 objects with significant (>3 σ ) variability in at least one band while dust emission can be only fitted in8 objects. Their peak dust luminosity is around 10 − erg s − , corresponding to a dust covering factor f c ∼ .
01 at scale of sub-parsec. The only exception is the disputed source ASASSN-15lh, which shows anultra-high dust luminosity ( ∼ . erg s − ) and make its nature even elusive. Other non-detected objects showeven lower f c , which could be one more order of magnitude lower. The derived f c is generally much smaller thanthose of dusty tori in active galactic nuclei (AGNs), suggesting either a dearth of dust or a geometrically thin andflat disk in the vicinity of SMBHs. Our results also indicate that the optical TDE sample (post-starburst galaxiesoverrepresented) is seriously biased to events with little dust at sub-pc scale while TDEs in dusty star-formingsystems could be more efficiently unveiled by IR echoes. Keywords: galaxies: sample — galaxies: nuclei — galaxies: ISM INTRODUCTIONSupermassive black holes (SMBHs) are ubiquitous in thecenters of galaxies with massive bulges. Moreover, the tight-ness of the BH-bulge mass relationship hints a symbioticconnection between the formation and growth of BHs andgalaxy spheroids (Kormendy & Ho 2013). The SMBHs ac-cumulate their tremendous mass (10 − M (cid:12) ) by accret-ing gas through the phase of active galactic nucleus (AGNs)while they are mostly quiescent in the local universe. A fun-damental question has been raised why some SMBHs are ac-tive but the majority of remaining are not (e.g., Alexander &Hickox 2012). Observational constrains on the AGN trigger-ing mechanisms remain elusive with only sparse evidence atgalactic scales (Storchi-Bergmann & Schnorr-Müller 2019).As the accreting material, the interstellar medium (ISM) atdifferent scales might give important clues to the underlying Corresponding author: Ning [email protected] mechanism. It is found that Seyfert AGNs generally reside inhost galaxies with a younger stellar population than quiescentgalaxies, confirming that an abundant fuel supply is availablein kiloparsec (kpc) scale (e.g., Kauffmann et al. 2003). Fur-ther efforts are devoted to explore their differences at smallerscale, such as finding a factor of four difference in gas massbetween Seyfert and quiescent galaxies within radius of 100pc (Hicks et al. 2013). Nevertheless, a clear picture of AGNtriggering is unachievable without going deep into the prox-imity of SMBHs, that is down to pc -scale under the gravita-tional influence of SMBHs.The structure of AGNs is cognized under the scheme ofunified model (Antonucci 1993), in which the equatorial op-tically thick torus at pc -scale lays the foundation of unifi-cation and has bridged the scale of accretion disk and theirhost galaxy. Besides, the dust along the polar direction mayalso exist in some AGNs suggested by interferometric ob-servations (Hönig et al. 2012; see also Lyu & Rieke 2018),leading to the dusty wind model (Hönig et al. 2013; Hönig& Kishimoto 2017). In contrast, the study of pc -scale envi-ronment of normal galaxies is much more challenging with- a r X i v : . [ a s t r o - ph . GA ] F e b J IANG ET AL .out the illumination from central engines. The far-IR imag-ing of our Milky Way center has uncovered a circumnuclearring centered on Sgr A (cid:63) with thickness and radius of 0.34 pcand 1.4 pc, respectively (Lau et al. 2013; see also Latvakoskiet al. 1999 for an agreed result). However, similar map isimpossible for more distant galaxies due to poor resolution.The groundbreaking instrument GRAVITY mounted on thevery large telescope has achieved mili-arcsec resolution in K -band, but it only applies to K -band luminous sources andthus barely nearby AGNs have been successfully observed(GRAVITY Collaboration et al. 2020a,b).Nevertheless, the gas and dust in the vicinity of inactiveSMBHs have another possibility to be lighted up temporar-ily by tidal disruption events (TDEs), which happens whena star occasionally passes within the tidal radius of SMBH.Part of the disrupted stellar debris will be accreted by the BHand produce a flash of electromagnetic radiation peaked inUV or soft X-ray band, with a characteristic t − / decline ontimescale of months to years (Rees 1988; Phinney 1989). Ifthe local environment of a SMBH is dusty, the UV/opticalphotons from TDE will be unavoidably absorbed and re-processed into the infrared (IR) band, giving rise to a so-called IR echo. Lu et al. (2016) has performed a 1-D radia-tive transfer model and proved that the dust emission peaksat mid-IR (MIR) with typical luminosity between 10 and10 erg s − depending on the dust covering factor. Imme-diately after the prediction, Jiang et al. (2016) achieved thedetection of IR echo of dust at scale of ∼ . pc in ASASSN-14li. At almost the same time, van Velzen et al. (2016)have claimed another two echoes (PTF-09ge and PTF-09axc)and derived a dust covering factor of ∼
1% for both events.Therefore, IR echoes of TDEs offer us a new and powerfulmeans to probe the dust around the BH down to sub-pc scale.The technique of IR echo is subject to the occurrence ofTDEs. The TDE event rate is yet known to be as low as10 − − − / galaxy / year (Wang & Merritt 2004; Stone &Metzger 2016) and thus makes the discovered events quiterare until recently. The number of TDEs has been grow-ing rapidly in the past decade thanks to the booming wide-field optical surveys, such as PanSTARRS, PTF, and ASAS-SN (Komossa 2015; van Velzen et al. 2020a). Particularly,the ZTF survey (Graham et al. 2019) since 2018 has madeTDE enter into a new era of population studies (van Velzenet al. 2021). As of the end of 2019, approximately 30 TDEshave been discovered at optical bands (see Table 1 of vanVelzen et al. 2020a). On the other side, the detection of IRechoes all have used the archival data of Wide-field InfraredSurvey Explorer ( WISE ; Wright et al. 2010) and Near-EarthObject
WISE
Reactivation mission (
NEOWISE -R; Mainzeret al. 2014). Actually, the dataset has provided multi-epochMIR date with time coverage matching with almost all opti-cal TDEs. There is however still no statistical study to date. Giving the increased number of optical TDEs in the past fewyears and available IR data, it is the perfect time to perform acomprehensive study. We assume a cosmology with H = 70km s − Mpc − , Ω m = 0 .
3, and Ω Λ = 0 . SAMPLE AND DATA2.1.
TDE Sample
The TDE candidates studied in this work are primarily col-lected from van Velzen et al. (2020a), which has reviewedall TDEs discovered in optical band up to 2019. We haveonly selected optical TDEs since they usually possess well-sampled multi-wavelength light curves with wide time spanwhich is important for us to get the knowledge of the basicproperties of these events, such as the peak time and luminos-ity. Furthermore, we only investigate events found between2009 and 2018 to ensure available MIR data within one yearafter the TDE, since the
WISE project starts from early 2010and its public data goes on to the end of 2019. The cut re-sults in 22 sources. In addition, we have also included an-other controversial TDE candidate ASASSN-15lh (Leloudaset al. 2016; Krühler et al. 2018) which was first reported asthe most ever luminous supernova (Dong et al. 2016). Wetake it into consideration in hope of gaining some new cluesfrom the IR variability. Thus, our final sample has 23 objectsin total (see their information in Table 1).2.2.
MIR Data
The
WISE has conducted a full-sky imaging survey in fourbroad MIR bandpass filters centered at 3.4, 4.6, 12 and 22 µ m(labeled W1-W4) from 2010 February to August (Wright etal. 2010). The solid hydrogen cryogen used to cool the W3and W4 instrumentation was depleted later and it was placedin hibernation in 2011 February. WISE was reactivated andrenamed
NEOWISE -R since 2013 October, using only W1and W2, to hunt for asteroids that could pose as impact haz-ard to the Earth (Mainzer et al. 2014). The
WISE scans a spe-cific sky area every half year and average 12 times of singleexposures have been taken within each epoch (typically oneday). As of now, the
WISE and its successor
NEOWISE sur-veys have provided us a public dataset from 2010 Februaryto 2019 December, which contains 14-15 epochs of obser-vations for each TDE. Therefore, the observing schedule of
WISE is in excellent overlap with the discovery period of theoptical TDEs in our sample (2009-2018).Our previous works have shown that the IR echoes ofTDEs are detectable on time scales of months to years whilethe variability within each epoch is negligible (Jiang et al.2016, 2017, 2019; Dou et al. 2016, 2017), so the origi-nal single-exposure photometry have been simply binned inthose works. However, it is not accurate enough to detectweak variability or put clear upper limit of the non-detectionsources. In order to acquire more accurate measurements,R
ECHO OF
TDE 3
Epoch 11iPTF-16fnl W1Epoch 1 Difference W2
Figure 1.
We show iPTF-16fnl as an example of image subtraction. The IR variability of this TDE are invisible from the original
WISE lightcurves while they are robustly detected with PSF photometrry in the difference images (epoch 1 as reference). particularly for TDEs with weak echoes, We choose toperform photometry on the time-resolved
WISE / NEOWISE
Coadds. The coadds have stacked the individual exposureswithin typically ∼ −− that is, one coadd every six months at agiven position on the sky (Meisner et al. 2018) . In addi-tion, the associated noise and mask images have been alsogenerated during the process. Therefore, it provides us aconvenient dataset which is very suitable for study the long-timescale MIR variability. ANALYSIS AND RESULTS3.1.
Variability Detection by Image Subtraction
We try to detect variability using the standard image sub-traction procedure
HOTPANTS (Becker 2015) . The imagesat latest epoch are taken as the references to be subtracted forTDEs discovered before 2013, otherwise the first epoch im-ages are adopted. Then we begin to perform PSF photometryon the difference images using the IDL routine
FASTPHOT (Béthermin et al. 2010). The PSF models specifically con-structed for coadd images (Meisner & Schlafly 2019) havebeen used as the input PSF images during the measurement. Website link: https://portal.nersc.gov/project/cosmo/temp/ameisner/neo6 https://github.com/acbecker/hotpants https://github.com/legacysurvey/unwise_psf Since the single-epoch reference image we used above canbe slightly offset from the real quiescent level from the hostgalaxy emission, we begin to estimate the offset by averagingfluxes of epochs at least 180 days before the optical peak forTDEs after 2013, or at least 1500 days after optical peak forTDEs before 2013. Then we corrected the offset and addedits error to the fluxes of difference images. Finally, we haveobtained the light curves with background (host) emissionsubtracted (see Figure 2).We consider the flux at certain epoch with signal to noiseratio (S/N) higher than 3 as a robust detection of variability. Ifnone epoch satisfies the condition, we put a 3 σ upper limit ofthe fluxes in which the σ is determined by the mean errors ofall epochs. According to this criterion, 11 TDEs show vari-ability in either W1 or W2 band at one or multiple epochs.Among them, we note that the IR echoes of ASASSN-14li atthe first two epochs has been reported by Jiang et al. (2016).PTF-09ge and PTF-09axc only shows 3 σ signal in W1 bandbut not in W2 band, that is also consistent with the resultsgiven by van Velzen et al. (2016). All of the measurementsare presented in Table 2.3.2. Dust Emission
After the detection of IR variability, the dust emission canbe then estimated. Although the extrapolated emission of theUV-optical blackbody should be generally weak in the MIRbands, it could be not negligible at the very early stage of J
IANG ET AL . Table 1.
Sample of Optical TDEs
ID Name IAU Name R.A. DEC. z MJD peak log L bb log< T BB > log t p log M BH log M (cid:63) erg/s K Day M (cid:12) M (cid:12) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)1 PTF-09axc ... 14:53:13.08 +22:14:32.3 0.1146 55016 43 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . . + . − . . + . − . ... ... 8 . + . − . . + . − .
10 iPTF-15af ... 08:48:28.13 +22:03:33.4 0.0790 57061 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
11 iPTF-16axa ... 17:03:34.34 +30:35:36.6 0.1080 57523 ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
12 iPTF-16fnl ... 00:29:57.01 +32:53:37.2 0.0163 57626 43 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
13 OGLE16aaa ... 01:07:20.81 -64:16:21.4 0.1655 57414 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
14 PS17dhz AT2017eqx 22:26:48.37 +17:08:52.4 0.1089 57910 43 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
15 PS18kh AT2018zr 07:56:54.54 +34:15:43.6 0.0710 58180 43 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
16 ASASSN-18pg AT2018dyb 16:10:58.77 -60:55:23.2 0.0180 58346 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
17 ASASSN-18ul AT2018fyk 22:50:16.13 -44:51:52.4 0.0590 58317 ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
18 ASASSN-18zj AT2018hyz 10:06:50.87 +01:41:34.1 0.0457 58427 ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
19 ZTF19aabbnzo AT2018lna 07:03:18.65 +23:01:44.7 0.0910 58508 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
20 ZTF18aahqkbt AT2018bsi 08:15:26.62 +45:35:32.0 0.0510 58217 ∗ . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
21 ZTF18abxftqm AT2018hco 01:07:33.64 +23:28:34.3 0.0880 58401 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
22 ZTF18acaqdaa AT2018iih 17:28:03.93 +30:41:31.4 0.2120 58442 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − .
23 ZTF18actaqdw AT2018lni 04:09:37.65 +73:53:41.7 0.1380 58460 44 . + . − . . + . − . . + . − . − . + . − . . + . − . . + . − . N OTE — Column (1): Object ID in this paper. Column (2): Discovery Name of the TDE. Column (3): IAU Name of the TDE. Column (4)-(5): RA and DEC of theTDE. Column (6): redshift of the TDE host galaxy. Column (7): MJD of the optical peak. The asterisks indicate the time of fist detection since these TDEs are onlydetected post-peak. Column (8): the UV-optical bolometric luminosity ( L bb ) the optical peak which is drawn from van Velzen et al. (2020a). Column (9): meanblackbody temperature ( T BB ) measured during the first 100 days post peak. Column (10): linear T BB change during the first year of observations. Column (10)-(11): p and t are the free parameters of a power-law decay ( L bb ∝ ( t / t ) p ). Column (12): black hole mass ( M BH ) derived from M BH - σ (cid:63) relation for target 1,2,3,4,6,7,10,11,12(Wevers et al. 2017) and 9 (Krühler et al. 2018); from model fitting for target 5,8,13 (Mockler et al. 2019); from M BH - M (cid:63) for other objects (Reines & Volonteri2015). Column (13): host stellar mass. The data from Column (7) to (11) are all drawn from van Velzen et al. (2020a) except for ASASSN-15lh. TDEs. For instance, the IR variability of ASASSN-18pg isdetected at almost the same time with the optical peak. Theexpected logarithmic luminosity at W1 and W2 band of theoptical blackbody ( T BB = 2 . × K ) is 41.09 and 40.69,that is only 0.54 and 0.65 dex lower than observations, re-spectively (see Figure 3).In order to minimize the contamination of optical black-body, we then try to subtract its contribution from the ob-served IR flux. We adopted the parameters listed in vanVelzen et al. (2020a) to characterize the optical luminosity: L bb ( t ) = L bb , peak (cid:18) t − t peak + t t (cid:19) p , (1)in which the relevant parameters can be found in Table 1.Here we assume that blackbody temperature ( T BB ) keeps con-stant to be the average of the first 100 days. Since ASASSN-15lh displays a non-monotonic declination with an exotic re-brightening feature in the UV-optical light curves (Leloudaset al. 2016), we have estimated its extension to the IR bandfrom the measured luminosity at corresponding WISE epochsdirectly. After flux correction, the dust temperature ( T dust ) can beinferred as below. f ν = 14 π d L (cid:90) a max a min N ( a )4 π a Q ν ( a ) π B ν ( T ) da (2)Following our previous works (e.g., Jiang et al. 2021), wesimply assume that the dust grains follow a MRN size distri-bution (Mathis et al. 1977; see also Draine & Lee 1984) as N ( a ) ∝ a − . with a min = 0 . µ m, a max = 10 µ m and an aver-age density of ρ = 2 .
7g cm − for silicate grains.Since PTF-09axc shows negative flux in W2 band at theepoch of 3 σ W1 detection, we have ignored it in the follow-ing analysis. In addition, the calculated T dust of some epochsare unreasonably higher (with large errors) than the allowedtemperature ( < − K ) suppressed by dust sublima-tion (Barvainis 1987; Mor & Netzer 2012). Their are mainlycaused by low S/N detection in W2 bands. For this reason,we have abandoned the epochs with measured T dust > K since their dust properties at corresponding epochs are ob-viously unreliable. Accompanied with T dust , the dust lumi-nosity ( L dust ) is obtained for the 8 TDEs with robust detec-R ECHO OF
TDE 5
Table 2.
IR emission of TDEs
ID Name Days f W f W2 log L W1 log L W2 log L dust T dust mJy mJy erg/s erg/s erg/s K(1) (2) (3) (4) (5) (6) (7) (8) (9)1 PTF-09axc 182 0.033 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
968 ASASSN-15lh 697 0.087 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
988 ASASSN-15lh 1153 0.076 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± OTE — Column (1): Object ID in this paper. Column (2): Discovery Name of the TDE. Column (3): Rest-frame days since the opticalpeak. Column (4): W1 flux in unit of mJy. Column (5): W2 flux in unit of mJy. Column (6): W1 luminosity. Column (7): W2luminosity. Column (8)-(9): Fitted dust luminosity and temperature. The contribution from the UV-optical components has beensubtracted off during the fitting.Only the epochs with 3 σ detection have been presented, while 3 σ upper limits are given for TDEs showing < σ detections at all epochs.We have not given the values of Column (8) and (9) for epochs with unreliable measurements ( T dust > K ). J IANG ET AL . Figure 2.
The W1 (3.4 µ m, blue) and W2 (4.6 µ m, red) light curves of TDEs. The fluxes (in unit of mJy) are measured with PSF photometry onthe difference images. The green dashed lines mark the time of the optical peak. We have highlighted the epochs with robust IR echo detections(dust emission measurements) in cyan shadow regions, and the epochs with only 3 σ W1 detection in grey shadow regions. R ECHO OF
TDE 7
Figure 2. continued tion in both W1 and W2 bands (see Table 2). The change of L dust with time is shown in Figure 4. The peak luminosity ismostly at orders of 10 − erg s − except for ASASSN-15lh (3 . × erg s − ), whose nature is still under hotlydebated (Dong et al. 2016; Leloudas et al. 2016).The spatial scale of dust can be roughly inferred from thetime lag between the peak dust emission ( L dust , peak ) and pri-mary optical-UV emission ( L BB , peak ). Apparently, the esti-mation have large errors due to poor cadence (half year) ofMIR light curves. As a conservative treatment, we put an er-ror of 180 days (observer-frame) to the MIR peak time. Onthe other hand, there are 7 TDEs which are detected only at the post-peak stage and their first detection epochs have beenadopted as alternatives of the peak time (marked as asterisksin Table 1). The errors introduced in this step might be mi-nor as the luminosity at the first detection ( ∼ erg s − )is comparable with the peak luminosity of other TDEs, indi-cating that the time of first detection is very close to theirreal peak. It is also confirmed by the model fitting ofASASSN-14ae, ASASSN-14li and ASASSN-15oi (Mockleret al. 2019). Consequently, the time lag errors are mainlydominated by the peak time of dust emission. The corre-sponding dust scale from lags are presented as ranges in Ta-ble 3. We can conclude that the dust revealed by IR echoes J IANG ET AL . Figure 3.
The SED of ASASSN-18pg at +0 days. The black line isthe blackbody spectrum determined from the optical-UV photome-try while the red circles denotes the observed luminosity at W1 andW2 band. The observed IR emission shows evident excess relativeto the UV-optical component.
Figure 4.
The fitted dust luminosity of the 8 TDEs with significantdetections of IR echoes. We show their evolution with time (therest-frame days since the optical peak). are all located at sub-pc, mostly at (cid:46) . pc , even if it is hardto measure precisely with available data.Another parameter widely adopted to characterize dustcontent is the covering factor of dust ( f c ). Here we try toestimate it by the ratio of peak dust emission and opticalemission, that is f c = L dust , peak / L BB , peak . The 8 objects with L dust , peak measurements show f c ∼ .
01 (see Table 3). Itneeds to be emphasized that f c of ASASSN-15lh is on thesame level albeit with much higher luminosity. We yet no- Figure 5.
The logarithmic dust covering factor (log f c ) as func-tion of redshift. The black solid circles represent the 8 TDEs withreliable detection in both W1 and W2 bands while the three withonly detection in W1 band are shown as green triangles. The non-detected 12 sources are plotted with open red circles (upper limits).As a comparison, we have also overplotted the two well sampledTDE candidates in AGNs, that is PS16dtm (Blanchard et al. 2017;Jiang et al. 2017) and PS1-10adi (Kankare et. 2017; Jiang et al.2019) with blue squares. The f c of PS16dtm is a lower limit sinceits MIR light curves is still rising. Both f c are apparently higherthan other TDEs in normal galaxies, which is consistent with theAGN torus. ticed that its MIR luminosity displays a much slower decaywith that of the latest epoch remains comparable with thepeak after 2 years. If we employ the energy ratio as an alter-native f c estimate of ASASSN-15lh, the integrated IR energy(1 . × erg) as of the end of 2019 yields out f c ∼ .
09 pro-vided a total optical energy ∼ (1 . − . × erg (Godoy-Rivera et al. 2017). The high IR luminosity of ASASSN-15lh is not only unusual in our TDE sample, but also muchhigher than other superluminous supernova (L.M.Sun et al.in preparation). Any future explanations of its nature mustalso account for the distinctive IR light curve successfully.Regarding the three with only reliable detection in W1band, we have made a very crude assumption of the T dust (1000 K ) and have got a very similar f c . At the end, thereare 12 objects left which have not been detected in W1 norin W2 band. We choose to just give the 3 σ upper limits of L dust derived from the W1 luminosity (upper limits) as wellas a fixed 1000 K T dust . The limits of L dust are mainly de-pendent on the redshift of TDEs, with closer TDEs havinglower upper limits (see Figure 5). Taking the nearest (amongthe non-detected) TDE ASASSN-14ae (z=0.0436) as an ex-ample, L dust <1.3 × erg s − , resulting in f c <1 . × − . Tosum up, optical TDEs show f c ∼ .
01 or even less, whichcan be as low as f c (cid:46) − . Because of the low dust content,R ECHO OF
TDE 9
Table 3.
Dust Scale and Covering Factor
ID Name t peak R d log L BB , peak log L dust , peak log f c Days pc erg/s erg/s(1) (2) (3) (4) (5) (6) colhead(7)3 PTF-09ge 200 0-0.31 44.04 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± OTE — Column (1): Object ID in this paper. Column (2): Discovery Name of theTDE. Column (3): The rest-frame days of the MIR peak since the optical peak (orfirst detection). Column (4): The range of the characteristic dust scale estimatedfrom the time lag between MIR and optical emission, in which the uncertaintiesare mainly caused by the poor cadence of MIR light curves. Column (5): Thepeak (or first detection) blackbody luminosity of the optical-UV emission. Col-umn (6): The peak dust luminosity with the UV-optical contribution subtracted off.Column (7): The dust covering factor estimated as f c = L dust , peak / L BB , peak . The L dust of the first 8 objects, whose variability detections are reliable in both W1 and W2bands, are calculated directly from the dust emission fitting. The 3 objects in themiddle, show >3 σ detection only in W1 band, thus we have naively assumed that T dust =1000 K, that is comparable with the average of the first 8 objects. We denotetheir f c with asterisks, emphasizing their large uncertainties and likely underesti-mated errors. Lastly, only 3 σ upper limits have been put for the rest 12 objectsbecause their S/N are too low to infer the dust properties. only nearest TDEs (z<0.1 except for ASASSN-15lh) showdetectable IR echoes with WISE images (see Figure 5). CONCLUSION AND DISCUSSIONThe pc-scale environment, namely the gas and dust, aroundSMBHs plays an important role in understanding the trig-gering mechanism of AGNs. Nevertheless, there is hithertono statistical comparison of the environments between activeand quiescent galaxies. The efforts are mainly hindered bythe unachievable resolving power. By means of IR echoes ofTDEs, we now have the great opportunity to take a snapshotof dust at sub-pc scale in normal galaxies (Lu et al. 2016;Jiang et al. 2016; van Velzen et al. 2016). The study pre-sented here has reviewed IR echoes of all optical TDEs dis-covered in the past decade (2009-2018) taking full advantage of available data resources. Confident (>3 σ ) IR variabilityhas been detected in 11 targets and dust emission has beenmeasured in 8 among them. The concerned sub-pc scale dustrevealed by IR echoes shows covering factor f c ∼ .
01, withthe caveat that other non-detected sources show likely evenlower f c . Our conclusion agrees nicely with the pilot studyof two TDEs (van Velzen et al. 2016) while with higher sta-tistical significance owing to the increased sample size. Itis noticeable that our sample has gone beyond z > . SMBHs are Quiescent due to Lack of Gas Supplies
The torus in AGNs can reprocess the UV-optical photonsfrom accretion to IR band and its covering factor has beencommonly inferred from the the spectral energy distribution(SED) decomposition of the primary and reprocessed emis-sion. Past works have shown that the torus covering factoris averagely close to one half (e.g., Fritz et al. 2006; Moret al. 2009; Roseboom et al. 2013). Consequently, TDEsin AGNs must produce associated IR outburst as a result ofdust echoes. This scenario is fully supported by the ubiq-uitously detected luminous (10 -10 erg s − ) IR echoes inAGN TDEs (Dou et al. 2017; Jiang et al. 2017, 2019; Mattilaet al. 2018). Furthermore, the f c derived from them is alsogenerally consistent with the traditional SED fitting method(Jiang et al. 2019; see also Figure 5).The optical TDEs considered in this work all happened ininactive galaxies (including LINERs). They all show weak ornon-detected IR echoes with f c (cid:46) .
01, that is more than oneorder of magnitude lower than AGNs, corroborating the con-clusion given in van Velzen et al. (2016). It implies that eitherthe pc-scale dust is quite sparse or it is concentrated on a geo-metrically thin and flat disk. In any case, such few dust is def-initely not sufficient to form a standard torus. In comparison,the f c of the circumnuclear ring in the Galactic center is 0.12inferred from parameters given by Lau et al. (2013). We maythus conclude that the pc -scale dust of normal galaxies, rep-resented by the optical TDE hosts, is not only much less thanAGNs, but also less than the Milky Way. The selection effectis minimal here as they are all optically discovered withoutany prior IR information. In some theoretical models, AGNradiation pressure is necessarily involved to both produce thetorus toroidal structure and maintain its thick structure (e.g.,Krolik 2007; Wada 2012). One may wonder if torus remainsthere when an AGN is turned off. Our results hint that thetorus, which is first introduced into the AGN unification as a0 J IANG ET AL .toy model, may only exist in AGNs but not normal galaxies.Likewise, the dusty wind in the polar direction should be alsoabsent when the SMBH is inactive, otherwise the polar dustwill also responds to the TDE as an notable IR echo (e.g.,Mattila et al. 2018).In other words, SMBHs are dormant probably because ofa shortage of gas in the vicinity instead of any formidableforce to prevent gas flow to the BH. AGNs seemingly to beeasily triggered as long as the ambient pc-scale gas is rich.However, it is not the whole story. Apart from the TDEs an-alyzed in this work, which are captured by optical transientsurveys, there is another class of TDE candidates selectedby transient coronal line emitters and dust IR echoes (Wanget al. 2012, 2018; Yang et al. 2013; Dou et al. 2016). Theyimply that SMBHs can be quiescent even if they lurk in ISM-rich environments. Albeit we are aware of a selection effectof this technique itself, it is worthwhile to further explore ifthe circumambient gas runs into a stone wall of losing angu-lar momentum or they are on the eve of AGN phase (turn-onAGNs, e.g., Gezari et al. 2017; Yan et al. 2019) in the future.4.2.
Implications to Demography of TDE Hosts
One of the most puzzling open questions in TDE field isthat optical TDEs show an unexpected preference in post-starburst (or so-called E+A) galaxies, with the rate elevatedby a factor of ∼
100 (e.g. Arcavi et al. 2014; French etal. 2016, 2020b). Scenarios which may contribute to therate enhancement, such as SMBH binaries (Chen et al. 2009,Coughlin et al. 2019), central stellar overdensity (French etal. 2020a) and radial velocity anisotropy (Stone et al. 2018)have been proposed out. However, those scenarios can notwell address why TDEs are absent in galaxies with occurrentintense star formation (Guillochon 2017).The serendipitous discoveries of obscured TDEs in ultra-luminous infrared galaxies (ULIRGs) by IR echoes indicatepromisingly that the absence of TDEs in star-forming (SF)systems is at least partly due to dust attenuation (Tadhunteret al. 2017; Mattila et al. 2018; Kool et al. 2020; see also Sunet al. 2020). Actually, the TDE event rate of ULIRGs is es-timated to be even higher than post-starburst galaxies, whichcould be as high as ∼ − gal − yr − (Tadhunter et al. 2017,Kool et al. 2020). Aside from ULIRGs, ordinary SF galaxieslying at the main sequence can also contain plenty of dust inthe galactic nucleus. A recent modeling of optical TDE de-tections in surveys indeed suggest that the dust obscuration iscrucial for suppressing the TDE detection rate in SF galaxieswhile the unusual preference for post-starburst hosts can not be entirely explained (Roth et al. 2020). We caution that theirestimate of extinction based on Balmer decrement have twocaveats. First, the dust distribution is usually not uniform butthe extinction of TDEs is only dependent on the dust alongthe line of sight. Second, we still lack the knowledge of dustextinction in the galactic nucleus, which is likely differentfrom the SF regions. Thus further study on the impact ofdust extinction must take the caveats into consideration.Jiang et al. (2021) has performed a blind search of MIRoutburst in nearby galaxies and has yielded out a consider-able number of TDE candidates in SF galaxies. Their peakMIR luminosity (10 − erg s − ) is much higher thanthe dust echoes revealed in optical TDEs. This work andother progresses of TDE search by means of dust echoessuggest that the IR band is an efficient wavelength to un-veil TDEs embedded in dusty environment. In contrast, theoptical search is only prone to uncover TDEs in SMBHswith very low dust covering factor ( (cid:46) . Wide-field Infrared Survey Ex-plorer , which is a joint project of the University of California,Los Angeles, and the Jet Propulsion Laboratory/CaliforniaInstitute of Technology, funded by the National Aeronauticsand Space Administration. This research also makes use ofdata products from
NEOWISE-R , which is a project of theJet Propulsion Laboratory/California Institute of Technology,funded by the Planetary Science Division of the NationalAeronautics and Space Administration.REFERENCES
Alexander, D. M., & Hickox, R. C. 2012, New Astronomy, 56, 93Antonucci, R. 1993, ARA&A, 31, 473Arcavi, I., Gal-Yam, A., Sullivan, M., et al. 2014, ApJ, 793, 38 Barvainis, R. 1987, ApJ, 320, 537Becker, A. 2015, Astrophysics Source Code Library. ascl:1504.004Béthermin, M., Dole, H., Cousin, M., et al. 2010, A&A, 516, A43 R ECHO OF