SWIFT J2058.4+0516: DISCOVERY OF A POSSIBLE SECOND RELATIVISTIC TIDAL DISRUPTION FLARE?

Sw J2058+05 was discovered by the Burst Alert Telescope (BAT; Barthelmy et al. 2005) on the Swift satellite (Gehrels et al. 2004) as part of the hard X-ray monitor's automated transient search. It first reached the 6σ discovery threshold in a four-day integration covering the time period 2011 May 17–20, with a 15–50 keV count rate of 0.0044 ± 0.0006 counts s⁻¹ cm⁻² (Krimm et al. 2011).¹³ No significant emission was detected from this location after 2011 June 2. The resulting BAT (15–50 keV) light curve is plotted in the top panel of Figure 1.

Zoom In Zoom Out Reset image size

The observed BAT count rate is too low to measure statistically significant variations on timescales shorter than 4 d or to constrain the hard X-ray spectrum. A search of transient monitor data on the same 4 d timescale back to 2005 February shows no previous activity from the source with a 3σ count rate limit of <0.003 counts s⁻¹ cm⁻².

To confirm this discovery, we requested a Swift target-of-opportunity (ToO) observation, which began at 21:56 on 2011 May 28. Data obtained by the X-ray Telescope (XRT; Burrows et al. 2005) were reduced using the online analysis tools of Evans et al. (2009), and XRT spectra were fitted using xspec12.6.¹⁴ After correcting the XRT astrometry following Goad et al. (2008), we identified a bright point source at (J2000.0) coordinates $\alpha = 20^{\mathrm{h}} 58^{\mathrm{m}} 19\mbox{$.\!\!^{\mathrm s}$}85$, δ = +05°13'330, with a 90% localization radius of 17. The XRT continued to monitor Sw J2058+05; the resulting 0.3–10 keV light curve is plotted in the middle panel of Figure 1.

Superimposed on the secular decline (reasonably well described by a power law: L_X∝t^−2.2), the X-ray light curve shows some degree of variability on relatively short timescales (compared to the time since discovery). The most significant flaring is detected on 2011 June 2, with a change in flux of a factor of 1.5 occurring on a timescale of δt ≈ 10⁴ s. No significant variability is detected on shorter timescales; however, given the signal-to-noise ratio, we are not sensitive to similar variations (i.e., factor of two) on timescales shorter than ∼10³ s.

The average photon-counting mode spectrum is reasonably well described by an absorbed power law with photon index Γ = 1.61 ± 0.12 and N_{H, host} = (2.6 ± 1.6) × 10²¹ cm⁻² (χ² = 65.95 for 54 degrees of freedom, dof; see Butler & Kocevski 2007 for details of the spectral analysis). Using this time-averaged spectrum, we find that the peak (unabsorbed) 0.3–10 keV flux is f_X ≈ 7.9 × 10⁻¹¹ erg cm⁻² s⁻¹. The requirement for absorption in excess of the Galactic value (N_{H, Gal} = 6.5 × 10²⁰; Kalberla et al. 2005) is only significant at the 3.2σ level. Interestingly, the time-resolved hardness ratio is inversely correlated with the source flux (Figure 1), a behavior commonly seen in Galactic black hole binaries (e.g., Remillard & McClintock 2006).

To improve the astrometric precision of the X-ray localization, we obtained ToO observations of Sw J2058+05 with the High Resolution Camera (HRC; Murray et al. 1997) on the Chandra X-Ray Observatory. Observations began at 14:51 on 2011 June 22, for a total live exposure of 5.2 ks. Sw J2058+05 is well detected with (J2000.0) coordinates $\alpha = 20^{\mathrm{h}} 58^{\mathrm{m}} 19\mbox{$.\!\!^{\mathrm s}$}90$, δ = +05°13'320, and an astrometric uncertainty of 04 (Figure 2; the lack of additional point sources in the HRC image precludes us from improving the native Chandra astrometric precision). The X-ray flux shows weak evidence (1.5σ) for a decline over the course of the entire observation, but no statistically significant variability on shorter timescales.

Zoom In Zoom Out Reset image size

Finally, we note that the location of Sw J2058+05 was observed on 2000 March 16 by the Position Sensitive Proportional Counters on board ROSAT as part of the All-Sky Survey (Voges et al. 1999). No sources are detected in the field to a limit of f_X(0.1–2.4 keV) <10⁻¹³ erg cm⁻² s⁻¹, several orders of magnitude fainter than the observed outburst.

Following the discovery of the X-ray counterpart, we initiated a campaign to observe Sw J2058+05 in the ultraviolet (UV), optical, and near-infrared (NIR), with the Swift Ultraviolet-Optical Telescope (UVOT; Roming et al. 2005), the seven-channel imager GROND (Greiner et al. 2008) mounted on the 2.2 m telescope at La Silla Observatory, the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) mounted on the 10 m Keck I telescope, the Wide-Field Camera (WFCAM) on the United Kingdom Infrared Telescope (UKIRT), and the Auxiliary-port Camera (ACAM) on the 4.2 m William Herschel Telescope. All observations were reduced following standard procedures and photometrically calibrated with respect to the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006) in the NIR, Sloan Digital Sky Survey (SDSS; Abazajian et al. 2009) in the optical, and following Poole et al. (2008) for the UVOT.

The optical counterpart of Sw J2058+05 was first identified in our GROND imaging obtained on 2011 May 29 at (J2000.0) coordinates $\alpha = 20^{\mathrm{h}} 58^{\mathrm{m}} 19\mbox{$.\!\!^{\mathrm s}$}90$, δ = +05°13'322, with an astrometric uncertainty of 02 (radius; Rau et al. 2011a, 2011b). The counterpart appears point-like in all our images, with the tightest constraints provided by our Keck/LRIS images on 2011 Jun 29 (07 seeing; Figure 2). Since discovery, we detect statistically significant fading in the bluer filters, but the degree of optical variability is much smaller than that observed in the X-rays. A full listing of our photometry is provided in Table 1, while the light curves are plotted in the bottom panel of Figure 1.

Pre-outburst imaging of the location of Sw J2058+05 was obtained on 2005 September 28 as part of SDSS. No source consistent with the location of Sw J2058+05 appears in the SDSS catalogs. Manually inspecting the images, we calculate 3σ upper limits of u' > 21.9, g' > 23.5, r' > 23.5, i' > 23.0, and z' > 21.7 mag. The lack of a pre-existing counterpart in SDSS is indicative that the observed optical emission is dominated by transient light and not any underlying host galaxy.

We obtained a series of optical spectra of Sw J2058+05 on 2011 June 1 with the Deep Imaging Multi-Object Spectrograph (DEIMOS; Faber et al. 2003) mounted on the 10 m Keck II telescope. The instrument was configured with the 600 lines mm⁻¹ grating, providing spectral coverage over the region λ = 4500–9500 Å with a spectral resolution of 3.5 Å. The spectra were optimally extracted (Horne 1986), and the rectification and sky subtraction were performed following the procedure described by Kelson (2003). The slit was oriented at the parallactic angle to minimize losses due to atmospheric dispersion (Filippenko 1982).

A portion of the resulting spectrum is plotted in Figure 3. Superimposed on a relatively blue continuum, we identify several strong (rest-frame equivalent width W_r ≳ 1 Å) absorption features corresponding to Mg ii λλ2796, 2803, Fe ii λ2600, Fe ii λ2587, Fe ii λ2383, and Fe ii λ2344 at z = 1.1853 ± 0.0004. Given the U-band detection (z ≲ 2 assuming λ_Lyα ≲ λ_U), we assume that this redshift corresponds to the host galaxy of Sw J2058+05. No other significant features are detected, either in emission or absorption, over the observed bandpass. Specifically, we limit the flux from [O ii] λ3727 to be f < 3 × 10⁻¹⁷ erg cm⁻² s⁻¹ (3σ).

Zoom In Zoom Out Reset image size

A second spectrum was obtained with Keck/LRIS on 2011 June 29, covering the wavelength range 3500–9500 Å. We find no significant evolution in either the shape of the continuum or the observed absorption features over this time period.

We observed the location of Sw J2058+05 with the National Radio Astronomy Observatory's Karl G. Jansky Very Large Array (VLA; Perley et al. 2011)¹⁵ on 2011 June 29.36 and 2011 July 19.39. Observations were conducted in the C band (4–8 GHz) on June 29 and in the K (8–12 GHz) and X (18–26 GHz) bands on July 19. The array was in the A configuration for both epochs, and data were processed using standard routines in the AIPS environment.

Within the EVLA field of view, we detect a single point source with (J2000.0) coordinates $\alpha = 20^{\mathrm{h}} 58^{\mathrm{m}} 19\mbox{$.\!\!^{\mathrm s}$}898$, δ = +05°13'3225, with an (absolute) astrometric uncertainty of 50 mas. We report the following flux densities: on June 29, f_ν(ν = 4.5 GHz) = 0.88 ± 0.05 mJy and f_ν(ν = 7.9 GHz) = 0.84 ± 0.04 mJy; on July 19, f_ν(ν = 8.4 GHz) = 1.34 ± 0.04 mJy and f_ν(ν = 22.5 GHz) = 1.21 ± 0.15 mJy.

The EVLA localization is plotted as a cross in the right panel of Figure 2. Native EVLA astrometry is relative to the ICRS system, which we have also used for the optical (via the 2MASS point source catalog) and X-ray localizations (via the native Chandra pointing solution). All three positions are coincident, implying that the emission we are observing in different bandpasses all results from the same physical source.

This field had been observed previously by the VLA at 1.4 GHz, once on 1996 July 8 as part of the NRAO VLA Sky Survey (NVVS; Condon et al. 1998) and three times in 2009 (March 9, March 16, and May 9) as part of the Faint Images of the Radio Sky at Twenty-Centimeters (FIRST; Becker et al. 1995) survey. We inspected the images from these surveys and found no source at the position of Sw J2058+05, with 3σ limits of f_ν < 1.5 mJy (NVSS) and 0.38 mJy (FIRST).

In Figure 4, we plot the X-ray and optical luminosity of Sw J2058+05 at an (observer-frame) time of Δt ≈ 12 days after discovery. For comparison, we also plot analogous measurements for a sample of long-duration GRB X-ray afterglows (at the same epoch post-burst), as well as nearby AGNs, more distant, luminous quasars from SDSS, and some of the most dramatically variable blazars observed while in outburst. With an isotropic X-ray luminosity L_{X, iso} ≈ 3 × 10⁴⁷ erg s⁻¹, Sw J2058+05 is much too luminous at this late time to result from a GRB. Yet with an absolute magnitude of M_U ≈ −22.7 mag, the observed optical emission is several orders of magnitude underluminous compared with the brightest X-ray quasars and blazars. Even compared with simultaneous X-ray and optical observations of the strongest high-energy flares from some of the best-known blazars (e.g., Markarian 501, Pian et al. 1998; 3C 279, Wehrle et al. 1998; Markarian 421, Fossati et al. 2008), the ratio of X-ray to optical luminosity appears to be quite unique. However, Sw J2058+05 occupies a nearly identical region of this phase space as Sw J1644+57.

Zoom In Zoom Out Reset image size

This similarity is further reinforced when adding the radio observations to the broadband spectral energy distribution (SED). In Figure 5, we plot the SED of Sw J2058+05 at Δt ≈ 43 days after discovery, compared with analogous measurements for a sample of blazars, the most dramatically variable class of AGNs (and also known sources of relativistic jets; Urry & Padovani 1995; see Section 3.2). Blazars exhibit a well-defined luminosity "sequence" whereby the lower-frequency maximum of their double-peaked SED occurs at lower energies for more luminous sources (Fossati et al. 1998). While the radio and optical emission from Sw J2058+05 are comparable to those seen in low-luminosity blazars, the X-ray emission outshines even the brightest sources of this class. On the other hand, the SED of Sw J1644+57, at a time of Δt ≈ 20 days, provides a good match to the observed properties of Sw J2058+05 (particularly after correcting for the large but uncertain dust extinction from Sw J1644+57).

Zoom In Zoom Out Reset image size

While the sequence plotted in Figure 5 may represent the ensemble properties of blazars as a class, individual sources may exhibit quite different behavior. During an outburst, blazar SEDs can undergo radical changes—for example, as a result of a rapid brightening in 1997, the low-energy (synchrotron) spectral peak in Markarian 501 shifted frequencies by approximately two orders of magnitude (Pian et al. 1998). We therefore also compare the observed SED of Sw J2058+05 with that of several high-energy outbursts from known blazars in Figure 6. Even when compared with these extreme cases, the ratio of the X-ray to optical flux observed in Sw J2058+05 stands out, being similar only to the broadband properties of Sw J1644+57.

Zoom In Zoom Out Reset image size

The sole remaining defining characteristic of Sw J1644+57 is its association with the nucleus of a nonactive galaxy (Levan et al. 2011; Zauderer et al. 2011). At the current time, the observed optical emission from Sw J2058+05 appears to be dominated by the transient, and we have no constraints on the location of the host nucleus from pre-outburst observations. Future high-resolution observations, when the transient light has faded, should be able to constrain the transient-nucleus offset. That said, the detection of strong absorption features indicates that the line of sight penetrates a dense region of the host galaxy (as in the case of some GRBs).

Finally, we utilize the optical spectrum to search for evidence of past AGN activity. Like blazars, NLS1 galaxies are another subclass of AGNs that can sometimes power relativistic jets (e.g., Abdo et al. 2009a, 2009b; Foschini et al. 2011; Section 3.2). Similarly, some NLS1 galaxies have been associated with luminous (L_X ≈ 10⁴⁴ erg s⁻¹) X-ray outbursts (e.g., Grupe et al. 1995; Brandt et al. 1995). In the case of Sw J1644+57, all of the detected emission lines were unresolved, and standard diagnostic diagrams revealed no evidence for ionization from a hidden (i.e., Compton thick) AGN (Levan et al. 2011; Zauderer et al. 2011). The larger redshift precludes such an analysis for Sw J2058+05, as the standard diagnostic lines do not fall in the optical bandpass.

However, we have compiled all of the available spectra from SDSS of known NLS1 galaxies (from the catalog of Véron-Cetty & Véron 2010) with z > 0.5 (478 objects) and used these to create an NLS1 template. This composite spectrum is plotted alongside that of Sw J2058+05 in Figure 3. The template exhibits strong, broad emission from C ii] λ2324, [Ne iv] λλ2422, 2424, and in particular Mg ii λλ2796, 2803. None of these emission lines are detected in our spectrum of Sw J2058+05. In fact, of the 478 known NLS1 galaxies in SDSS with z > 0.5, all but one have well-detected (>3σ) Mg ii in emission. Unless exceedingly rare, the optical spectrum of Sw J2058+05 suggests that it is not an NLS1 galaxy.

We have furthermore performed a similar analysis with all known BL Lac sources from Véron-Cetty & Véron (2010) at z > 0.5 in the SDSS archive (68 sources); the resulting median spectrum is shown in red in Figure 3. By definition, BL Lac objects lack the bright emission features present in the spectra of NLS1 galaxies, and therefore they more closely resemble the observed spectrum of Sw J2058+05. However, BL Lac spectra do typically exhibit Ca ii λλ3934, 3968 in absorption, which is lacking in Sw J2058+05, and not a single BL Lac in the SDSS database exhibits strong Mg ii absorption. While we cannot entirely rule out a new mode of variability in a blazar, the unique SED and optical spectrum distinguish Sw J2058+05 from known members of this class as well.

We have shown in the previous section that Sw J2058+05 does not appear to be consistent with any known class of AGNs (particularly blazars and NLS1). Based on the many similarities with Sw J1644+57, we can use the same arguments to derive the basic physical parameters for Sw J2058+05 (e.g., Bloom et al. 2011). From our NIR observations, we derive an upper limit on the host-galaxy absolute V-band magnitude of M_V ≳ −21 mag (we have used an S0 galaxy template from Kinney et al. 1996 for the K-correction based on the lack of emission features in the optical spectrum). Applying the black hole mass vs. bulge luminosity (Magorrian et al. 1998) correlation from Lauer et al. (2007) and assuming a bulge-dominated system, we limit the mass of the putative central black hole to be M_BH ≲ 2 × 10⁸ M_☉.

Separately, the shortest observed X-ray variability timescale can be used to constrain the central black hole mass using causality arguments (e.g., Campana et al. 2011). For an observer-frame variability timescale of δt ≲ 10⁴ s, we find M_BH ≲ 5 × 10⁸ M_☉. The most dramatic short timescale variability observed from Sw J1644+57 occurred at Δt ≲ 10 days (Bloom et al. 2011), before we commenced X-ray observations of Sw J2058+05. But even at late times, the X-ray light curve of Sw J2058+05 exhibits a degree of variability (changes of an order of magnitude on timescales of a few days) not seen in Sw J2058+05, suggesting an imperfect analogy between these two sources.

Finally, we can constrain the central mass using the black hole fundamental plane, a relationship between mass, X-ray luminosity, and radio luminosity (e.g., Gültekin et al. 2009). Application of this relation requires knowledge of the beaming fraction, for which we adopt a value of f_b ≡ (1 − cos θ) ≈ 10⁻² (see below). Following Miller & Gültekin (2011), we find M_BH ≈ 5 × 10⁷ M_☉; however, we caution that the scatter in this relationship is even larger than that found in the black hole mass versus bulge luminosity relation.

The derived black hole mass, M_BH ≲ 10⁸ M_☉, is of vital importance for two main reasons. First, the tidal disruption of a solar-type star will only occur outside the event horizon for M_BH ≲ 2 × 10⁸ M_☉. As such, if the origin of the accreting material is indeed the tidal disruption of a nondegenerate star (or more massive object), the mass of the central black hole in the host galaxy of Sw J2058+05 must be sufficiently small.¹⁶ Future observations of the host, when the optical transient light has faded, should provide a significantly improved estimate of M_BH.

Second, for M_BH ≲ 10⁸ M_☉, the observed X-ray emission exceeds the Eddington luminosity. In particular, the peak isotropic 0.3–10 keV X-ray luminosity of L_{X, iso} ≈ 3 × 10⁴⁷ erg s⁻¹ exceeds the Eddington value by more than an order of magnitude. Given the long-lived nature of the central engine, this suggests that the outflow is likely to be collimated with a beaming factor f_b ≲ 10⁻¹ (cf., Socrates 2012). An even larger degree of collimation is required to reconcile the observed rate of these outbursts with theoretical predictions for TDFs (Section 4).

More robust evidence for beaming is provided by the large radio luminosity. It is well known that the brightness temperature of an incoherent radio source cannot exceed 10¹¹ K for extended periods of time (Readhead 1994; Kulkarni et al. 1998). Utilizing this fact, we can derive a lower limit on the angular size of the radio-emitting region, θ ≳ 14 μ as. Assuming that the outflow commencement coincides with the hard X-ray discovery, this implies a mean expansion velocity of β ≡ v/c ⩾ 0.88 or a mildly relativistic expansion with Γ ⩾ 2.1. Furthermore, the radio emission will be collimated into a viewing angle θ ≲ 1/Γ due to relativistic beaming effects.

If we integrate over the X-ray light curve to date (2011 July 20), we find that Sw J2058+05 has emitted a total 0.3–10 keV fluence of S_X ≈ 1.0 × 10⁻⁴ erg cm⁻². If we assume this accounts for ∼1/3 of the broadband luminosity, the total isotropic energy release is E_iso ≈ 10⁵⁴ erg, comparable to that of Sw J1644+57 at a similar epoch. For a typical radiative efficiency for accretion power (η ≡ E_rad/Mc² ≲ 0.1), the tidal disruption of a solar-type star would require a significant beaming correction. If only a small fraction of the available energy is allocated to the observed relativistic component (≲ 1%), this would imply f_b ≲ 10⁻³.

In addition to the weaker degree of X-ray variability, we wish to draw attention to two other important differences between these sources. Unlike that case of Sw J1644+57, the observed UV/optical SED of Sw J2058+05 is quite blue. In fact, the rest-frame UV data for Sw J2058+05 can be well fit by a blackbody spectrum (Figure 6), with T_BB ≳ 6 × 10⁴ K, L_BB ≳ 10⁴⁵ erg s⁻¹, and R_BB ≳ 10 AU.¹⁷ The derived blackbody parameters are similar to the thermal emission observed from previous TDF candidates discovered at UV and optical wavelengths (Gezari et al. 2006, 2008, 2009; van Velzen et al. 2011; Cenko et al. 2012), and are strongly suggestive of an accretion disk origin for the optical light. In this sense, Sw J2058+05 may serve as a link between the bright, nonthermal X-rays observed in the relativistic TDF candidates, and the thermal disk emission from "classical" TDFs. Such comparisons were not possible for Sw J1644+57 due to dust in the host galaxy.

More importantly, the radio spectral index of Sw J2058+05 appears to be quite flat (f_ν∝ν⁰). This stands in stark contrast to Sw J1644+57, which was optically thick (f_ν∝ν^1.3) up to high frequencies in the days and weeks following discovery (Zauderer et al. 2011). This may simply reflect physical differences in the jet (e.g., microphysics) or its environment (e.g., density), as is suggested by the different extinction properties. However, it may also imply a more extended (and hence long-lived) radio source, which would be difficult to reconcile with a tidal disruption origin. Future high-resolution radio interferometry (i.e., VLBI) may be able to resolve this issue.