TIDAL DISRUPTION EVENTS PREFER UNUSUAL HOST GALAXIES

If a star passes close enough to a supermassive black hole that the tidal forces overcome the self-gravity of the star, the star will be destroyed in a tidal disruption event (TDE; Hills 1975). Such events are expected to generate an observable flare (Rees 1988; Evans & Kochanek 1989; Phinney 1989) if the tidal radius is greater than the Schwarzschild radius.

Real-time discoveries of TDE candidates have enabled extensive followup observations and classification. The first was Swift J1644 (Bloom et al. 2011; Burrows et al. 2011; Levan et al. 2011; Zauderer et al. 2011), displaying non-thermal emission in γ-rays, X-rays, and the radio. Two additional events had similar properties: Swift J2058 (Cenko et al. 2012b) and Swift J1112 (Brown et al. 2015). Hereafter we refer to these events as "high energy TDEs."

In parallel, a different class of transients were also identified as likely TDEs. The first was PS1-10jh (Gezari et al. 2012), which had thermal optical and UV emission but no observed X-rays. Since the discovery of PS1-10jh, eight objects with similar TDE spectral features have been found. (Arcavi et al. (2014, hereafter A14) discovered three in Palomar Transient Factory (PTF) data, PTF09axc, PTF09djl, and PTF09ge, grouping them together as a class with PS1-10jh, SDSS J0748 (identified in the Sloan Digital Sky Survey (SDSS); Wang et al. 2011), and ASASSN-14ae (Holoien et al. 2014). Recently, three additional members of this class have been discovered: ASASSN-14li (Holoien et al. 2016), ASASSN-15oi (Brimacombe et al. 2015), and PTF15af (in the galaxy SDSS J084828.13+220333.4; N. Blagorodnova et al. 2016, in preparation). This optical/UV-selected class of transients all display hot blackbody ($\sim \mathrm{few}\cdot {10}^{4}$K) emission and several-month-long smooth light curves peaking at an absolute optical magnitude of $\sim -20$. They are all located in the centers of their host galaxies. Their clear broad H and/or He emission lines (A14) cleanly distinguish them from other transient events. Hereafter we refer to these events as "optical/UV TDEs."

Curiously, the host galaxy spectra of these eight⁴ events show Balmer line absorption, and all but SDSS J0748 have weak or no emission lines. This combination of spectral features indicates low levels of current star formation, yet substantial star formation in the last ∼Gyr, long enough ago for the ionizing O and B stars to have evolved away, but recently enough for A stars to dominate the stellar light. Galaxies with these spectral features are called Balmer-strong, Hδ-strong, E+A, k+a, or a+k galaxies, depending on the strength of the Balmer absorption and the authors' preference (Dressler & Gunn 1983; Couch & Sharples 1987; Dressler et al. 1999). Many are consistent with a post-starburst star formation history and galaxy–galaxy merger origin, and are likely in transition between star-forming spirals and passive early-type galaxies (Zabludoff et al. 1996; Yang et al. 2004, 2008).

Similarly, the host galaxy of the high energy TDE candidate Swift J1644 was reported to have significant Balmer absorption with a low current star formation rate (SFR) predicted from its Hα flux (Levan et al. 2011). Yoon et al. (2015) find a young $\lt 1\;{\rm{Gyr}}$ stellar population, suggesting a recent starburst. The two other known high energy TDE candidates do not yet have host galaxy spectra covering the full Balmer series.

The clear preference of optical/UV TDE candidates and at least one high energy TDE candidate for rare Balmer-strong galaxies has important implications for the mechanisms driving TDE rates. Here we examine the statistics of galaxies like the TDE candidate hosts to quantify the TDE-rate enhancement in such environments.

2.1. Quiescent Balmer-strong Classification

2.2. Data

Optical spectroscopy was obtained for the eight host galaxies of the optical/UV TDE candidates from A14, Yang et al. (2013), and the SDSS, and for the host of high energy TDE candidate Swift J1644 from Levan et al. (2011). We plot these host galaxy spectra in Figure 1, describing their features in Table 1.

Zoom In Zoom Out Reset image size

Table 1.  TDE Host Properties

TDE Host	Hα EWa	H${\delta }_{{\rm{A}}}$	z	Mrb	Slit Width	Data Source
	(Å)	(Å)		mag	(arcsec (kpc))
SDSS J0748	−11.36 ± 1.00	1.20 ± 0.81	0.0615	−20.13 ± 0.02	1.0 (1.2)	Yang et al. (2013)
ASASSN14ae	−0.68 ± 0.40	3.37 ± 0.79	0.0436	−19.75 ± 0.02	3.0 (2.6)	SDSS
ASASSN14li	−0.59 ± 0.53	5.71 ± 0.61	0.02058	−19.20 ± 0.02	3.0 (1.3)	SDSS
PTF09axc	−1.07 ± 0.67	4.89 ± 0.36	0.1146	−20.55 ± 0.02	1.0 (2.1)	A14
PTF09djl	−0.26 ± 0.66	4.67 ± 0.49	0.184	−20.02 ± 0.03	1.0 (3.1)	A14
PTF09ge	−1.70 ± 0.75	0.33 ± 0.68	0.064	−20.23 ± 0.02	0.7 (0.9)	A14
PS1-10jh	−0.54 ± 0.65	1.68 ± 0.76	0.1696	−18.48 ± 0.05	1.0 (2.9)	A14
PTF15af	−1.65 ± 0.30	1.31 ± 1.91	0.0790	−20.20 ± 0.02	3.0 (4.5)	SDSS
Swift J1644	−2.50 ± 0.76	4.71 ± 1.06	0.3534	−18.44 ± 0.1	1.0 (5.0)	Levan et al. (2011)

Notes.

^aNegative values indicate emission. Hα EW values are corrected for stellar absorption. ^bAbsolute magnitudes (r band, no extinction correction) from SDSS (model_mag) for optical/UV TDE hosts, and from Levan et al. (2011) for Swift J1644. We assume ${H}_{0}=70$ km s⁻¹ Mpc⁻¹, ${{\rm{\Omega }}}_{{\rm{\Lambda }}}=0.7$, and ${{\rm{\Omega }}}_{m}=0.3$.

Download table as:  ASCIITypeset image

For each TDE candidate host galaxy, we calculate ${\rm{H}}{\delta }_{{\rm{A}}}$ as described above, and Hα EW using two pseudo-continuum bands (Blue: [6492.8–6532.8 Å]; Red: [6592.8–6632.8 Å]). We use stellar population models from Conroy et al. (2009) and Conroy & Gunn (2010) to fit archival UV and optical photometry (from SDSS, GALEX, and Swift UVOT) together with the optical spectral features. We model the star formation history as an old population plus a recent, exponentially-declining burst of star formation. The age since the recent burst, mass produced and duration of the recent burst, and dust extinction are free parameters (K. D. French et al. 2016 in preparation). From these models, we determine the stellar absorption correction to the Hα EW, which ranges from 1.5–2.5 Å.

To quantify the rarity of these TDE hosts, we compare their spectral features to the SDSS main galaxy spectroscopic sample (Strauss et al. 2002), defining a parent sample from DR10 (Aihara et al. 2011). To prevent severe aperture bias, we exclude galaxies with $z\lt 0.01$ to eliminate those that are very large on the sky relative to the 3'' diameter of the SDSS fibers. We also exclude galaxies with unreliable⁵ Hα EWs or median signal-to-noise values of less than 10 per pixel over the whole spectrum. Our final parent sample is composed of 591,736 galaxies. We use the Hα EWs (corrected for stellar absorption) and Lick ${\rm{H}}{\delta }_{{\rm{A}}}$ values from the MPA-JHU catalogs (Brinchmann et al. 2004). In the three cases where the TDE host galaxy spectrum is from the SDSS, our ${\rm{H}}{\delta }_{{\rm{A}}}$ and corrected Hα EW measurements are consistent with the MPA-JHU catalog values.

The scarcity of quiescent Balmer-strong galaxies makes it unlikely to see many TDE host galaxies in this class, were TDEs to occur in all types of galaxies. Our strictest selection cut, ${\rm{H}}{\delta }_{{\rm{A}}}$ − σ(H${\delta }_{{\rm{A}}}$) > 4 Å and Hα EW < 3 Å (emission), on the SDSS produces a sub-sample of 1207 galaxies (0.20%). Of the optical/UV TDE host galaxies studied here, three (ASASSN14li, PTF09axc, PTF09djl) of the eight satisfy these cuts. Quiescent Balmer-strong galaxies are thus over-represented in the TDE host galaxy sample by a factor of ${190}_{-100}^{+115}$ times.⁶

The significance of these cuts is illustrated in Figure 2. SDSS galaxies fall into the "blue cloud," with Hα emission and moderate ${\rm{H}}{\delta }_{{\rm{A}}}$ absorption, and the "red sequence," with little-to-no Hα emission and low ${\rm{H}}{\delta }_{{\rm{A}}}$. Quiescent Balmer-strong galaxies lie in the "spur" to the lower right, with weak Hα emission and strong ${\rm{H}}{\delta }_{{\rm{A}}}$ absorption. Most of the TDE host galaxies lie along this spur.

Zoom In Zoom Out Reset image size

The example model tracks in Figure 2 show how galaxies with different star formation histories evolve. Two tracks follow a recent burst of star formation (an exponentially-declining model of short duration; $\tau \quad =$ 100 or 200 Myr) on top of an existing old stellar population. We again use the stellar population models from Conroy et al. (2009) and Conroy & Gunn (2010), assuming that the recent burst created 10% of the stellar mass of the galaxy. A shorter burst with the same mass fraction produces higher ${\rm{H}}{\delta }_{{\rm{A}}}$ absorption. Star formation declining over $\tau \gt 200\;{\rm{Myr}}$ does not generate high enough ${\rm{H}}{\delta }_{{\rm{A}}}$ to meet our strictest criterion. Thus, the three TDE host galaxies satisfying the strictest ${\rm{H}}{\delta }_{{\rm{A}}}$ cut have probably experienced a recent, strong, brief period of star formation, i.e., a true burst.

The ${\rm{H}}{\delta }_{{\rm{A}}}$ requirement can be relaxed to > 1.31 Å to encompass the three other optical/UV TDE host galaxies that also lie in the spur: ASASSN14ae, PTF15af, and PS1-10jh. This cut includes 13749 SDSS galaxies, or 2.3%, but six of the eight known TDE hosts. Thus, quiescent galaxies with at least moderately strong Balmer absorption are over-represented in the TDE host galaxy sample by a factor of ${33}_{-11}^{+7}$. This lower ${\rm{H}}{\delta }_{{\rm{A}}}$ cut allows several possible star formation histories, including star formation that has declined over several Gyr, starbursts that have ended and evolved past the strong ${\rm{H}}{\delta }_{{\rm{A}}}$ region, and constant star formation with a sudden truncation (e.g., Shioya et al. 2004). Therefore, it is less straightforward to interpret the recent histories of the three lower ${\rm{H}}{\delta }_{{\rm{A}}}$ galaxies; all have recently ended periods of star formation, but may not have experienced a strong starburst.

The two remaining TDE candidate host galaxies, SDSS J0748 and PTF09ge, do not lie in the spur of Figure 2. The TDE in SDSS J0748 was discovered in a search for high-ionization emission lines (Wang et al. 2011), not in a transient survey, a selection that might contribute to the distinct spectral properties of its host, which lies in the blue cloud of star-forming galaxies in Figure 2. In contrast, PTF09ge, found in the same manner as the other PTF A14 TDE candidates, has a host galaxy that lies closer to the red sequence of early-type galaxies.

Two of the optical/UV TDE host galaxies (PTF09djl and PS1-10jh) lie in the SDSS footprint, and have absolute magnitudes comparable to the other six host galaxies, but are too faint for the SDSS spectroscopic sample due to their slightly higher redshifts ($z=0.1696,0.184$). No significant galaxy evolution from $z\sim 0.2$ to 0.01 is expected (Snyder et al. 2011). Therefore, the quiescent Balmer-strong galaxy fraction should remain the same. To check this, we cut the parent sample to bound the redshifts of these two TDE hosts. The resulting TDE rate enhancements are within the 1σ errors of those quoted above.

The high energy TDE host galaxy Swift J1644 has some Hα emission (Figure 1), but it is weak enough to meet our criteria for being quiescent. The host galaxy spectrum is noisy around ${\rm{H}}{\delta }_{{\rm{A}}}$, and thus does not meet our strictest Balmer-strong cut of ${\rm{H}}{\delta }_{{\rm{A}}}$ − σ(H${\delta }_{{\rm{A}}}$) > 4 Å. However, stronger Hβ (EW $=\quad 8.2\pm 1.1$ Å) and Hγ (EW $=\quad 11.7\pm 2.4$ Å) absorption are reported by Levan et al. (2011). Because this galaxy is at a higher redshift (z = 0.3534) than the optical/UV TDEs, we constrain the SDSS comparison sample to $z\gt 0.3$ in this case. A Swift J1644-consistent cut on ${\rm{H}}{\delta }_{{\rm{A}}}$ − σ(H${\delta }_{{\rm{A}}}$) > 3.65 Å and Hα EW < 3 Å yields 10 out of 2287 galaxies, or 0.4%. Although we do not have full optical spectra for the other two high energy TDEs, it is significant that at least 1/3 of that sample lies within only 0.4% of the parameter space.

4.1. Preference of TDEs for Quiescent Balmer-strong Galaxies

4.2. Implications for the TDE Rate

We demonstrate the preference of TDEs to occur in quiescent galaxies with strong Balmer line absorption. Quiescent galaxies with the strongest Balmer absorption, ${\rm{H}}{\delta }_{{\rm{A}}}$ − σ(H${\delta }_{{\rm{A}}}$) > 4 Å, make up only 0.2% of local galaxies, yet host 3 of 8 optical/UV TDE candidates. A softer cut, ${\rm{H}}{\delta }_{{\rm{A}}}\gt 1.31$ Å, includes only 2.3% of local galaxies, but 6 of 8 optical/UV TDE host galaxies. The optical/UV TDE rates are thus enhanced by ${190}_{-100}^{+115}\times$ in the strongest ${\rm{H}}{\delta }_{{\rm{A}}}$ galaxies and by ${33}_{-11}^{+7}\times$ in galaxies with ${\rm{H}}{\delta }_{{\rm{A}}}\gt 1.31$ Å. Because of this preference, the corresponding rates of optical/UV TDEs are 1–3 × ${10}^{-3}$ yr⁻¹ per galaxy, and 2–4 × ${10}^{-4}$ yr⁻¹ per galaxy, respectively. As a result, we predict a lower optical/UV rate in normal star-forming and early-type galaxies, of 1–5 × ${10}^{-6}$ yr⁻¹ per galaxy. Even the one high energy TDE with a full measured optical spectrum, Swift J1644, lies in a galaxy with strong Balmer absorption and weak nebular line emission, which implies a $\gt 80\times$ enhancement in such hosts, and a link between the optical/UV and high energy TDE classes.

Why do TDEs prefer quiescent, Balmer-strong hosts? This type of galaxy has several properties that make it special and may suggest an answer. Many have had a recent galaxy–galaxy merger (Zabludoff et al. 1996), increasing the possibility of a black hole binary, perturbed stellar orbits, a spatially concentrated population of A stars, and/or an evolved population of easily disrupted A giants. The high incidence of LINER-like emission (Yan et al. 2006; Yang et al. 2006) may also play a role. In future work, we will explore these connections.

We thank A. Levan and A. de Ugarte Postigo for providing the digital version of their Swift J1644 host galaxy spectrum. We thank Yujin Yang for his contributions to the stellar population fitting routines. K.D.F. acknowledges support from NSF grant DGE-1143953, P.E.O., and the ARCS Phoenix Chapter and Burton Family. I.A. is grateful for support from the Karp Discovery Award. A.I.Z. and I.A. thank the Center for Cosmology and Particle Physics at NYU for its support. A.I.Z. acknowledges funding from NSF grant AST-0908280 and NASA grants ADP-NNX10AD476/ADP-NNX10AE88G. A.I.Z. thanks the John Simon Guggenheim Foundation for its support.

Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. The SDSS-III web site is http://www.sdss3.org/.