The Search for Tidal Disruption Flares with with GALEX

Suvi Gezari

California Institute of Technology

& The GALEX Team

California Institute of Technology,

Laboratoire Astrophysique de Marseille

Xi’an AGN 2006 -- October 21, 2006

The Search for Tidal Disruption Flares withThe Search for Tidal Disruption Flares with with GALEXwith GALEX

OutlineOutline

i. Capability of GALEX to Study Variability

ii. Tidal Disruption Flares: Theory and Observations

iii. Searching for Flares with GALEX

iv. Tidal Disruption Flare Discovered

v. Future Dedicated Time Domain Survey

Capabilities of GALEXCapabilities of GALEX

• Deep Imaging Survey, 80 deg2, >30 ksec (mlim ~ 25)• Observations obtained in 1.5 ksec eclipses• Time-tagged photon data (time resolution of 5 msec)• Large field of view (1.2 sq. deg.) and a large survey volume• Low sky background (source detection with 10 photons)• Simultaneous NUV (1750 - 2750 Å) and FUV (1350 - 1750 Å) imaging• Simultaneous R=100/200 NUV/FUV spectroscopic grism data

SDSSGALEX

A star will be disrupted when:Rp < RT ≈ R(MBH/M)1/3

Evans & Kochanek (1989)

Tidal Disruption EventsTidal Disruption Events

The bound fraction (< 0.5) of the stellar debris falls back onto the black hole, resulting in a luminous accretion flare.

Evans & Kochanek (1989)

t-5/3

Properties of a Tidal Disruption FlareProperties of a Tidal Disruption Flare

• For 106 - 108 M black holes, the stellar debris accretes in a thick disk (Ulmer 1999)

• Lflare≈LEdd=1.31045 M7 erg s-1

• Teff≈(LEdd/4RT2)1/4=3105 M7

1/12 K

• L(t)=(dM/dt)c2t-5/3

• dN/dt7/2MBH-1MBH

-1/4≈10-4 yr-1

(Wang & Merritt 2004)

Tidal disruption theory predicts rare but luminous flares that peak in the UV/X-ray domain, with decay timescales ~ months.

Why Search For Tidal Disruption Events?Why Search For Tidal Disruption Events?

• They are an unambiguous probe for supermassive black holes lurking in the nuclei of normal galaxies.

• The luminosity, temperature, and decay of the flare is dependent on the mass and spin of the black hole.

• They may contribute to black hole growth over cosmic times, and the faint end of the AGN luminosity function.

• Tidal disruption rates are sensitive to the structure and dynamics of the stellar galaxy nucleus.

Flares Detected by ROSATFlares Detected by ROSAT

The ROSAT All-Sky Survey (RASS) conducted in 1990-1991 was an excellent experiment to detect TDEs since it sampled 3x105

galaxies in the soft X-ray band (0.1 - 2.4 keV).

• Tbb = 6 - 12 x 105 K• Lx = 1042 - 1044 ergs s-1

• tflare ~ months• Event rate ≈ 1 x 10-5 yr-1 (Donley et al. 2002)

Properties of a tidal disruption event!

NLSy1

Sy1.9

non-active

Halpern, Gezari, & Komossa (2004)

Follow-up narrow-slit HST/STIS spectra confirmed the galaxies as non-active (Gezari et al. 2003).

Lflare/L10yr = 240

Lflare/L10yr = 6000

Lflare/L10yr = 1000

HST Chandra

STIS

STIS

• Narrow-line emission requires excitation by a persistent Seyfert nucleus.

• Seyfert nucleus in it inner 0.”1 was previously masked by H II regions in ground-based spectra

• The Chandra upper-limit on the nuclear X-ray luminosity is consistent with the predicted Lx from L(H) for LLAGNs, of ~ 9 x 1038 ergs s-1.

Li et al. (2002) modeled the event as the partial stripping of a low-mass star, or the disruption of a brown dwarf or giant planet.

Halpern, Gezari, & Komossa (2004)

NGC 5905

Gezari et al. (2003)

Search for Flares in the Deep Imaging SurveySearch for Flares in the Deep Imaging Survey

TOO observations with Chandra and Keck will probe the early-phase of decay of TDEs for the first time!

We take advantage of the UV sensitivity, temporal sampling, and large survey volume of GALEX.

Selection of Variable SourcesSelection of Variable Sources

is a measure of the dispersion in mag due to photometric errors

Identify the Host of the FlareIdentify the Host of the Flare

StarsQuasarsGalaxies

CFHTLS Colors & Morphology

Followed-up with TOO optical

spectroscopy

The most convincing candidates have inactive galaxy hosts!

Optically unresolvedo Optically resolvedx Xray detection Optically variable

Spectroscopic AGN

Archival Chandra ACIS detection in April 2002 with Lx ≈ 9.3x1042 ergs s-1

Presence of persistent Seyfert activity makes the tidal disruption scenario difficult to prove.

Jan 2006 MDM 2.4m spectrumL([O III]) = 9.4x1040 ergs s-1

Seyfert at z = 0.355!Some hosts show AGN activitySome hosts show AGN activity

AEGIS DEEP2 Keck Spectrum

Early-type galaxy with no evidence of Seyfert activity!

Confirmed Inactive Galaxy HostConfirmed Inactive Galaxy Host

Gezari et al. 2006, in press

UV Flare with tUV Flare with t-5/3-5/3 Decay Decay

t-5/3 The delay between tD and the peak of the flare implies MBH > 108 M.

tD

Gezari et al. 2006, in press

Properties of the Flare:T = 1 to 5 x 105 KL = 1044 to 1046 erg s-1

Macc > 0.3 Msun

In excellent agreement with theoretical predictions for a stellar disruption flare.

Simultaneous SED of the FlareSimultaneous SED of the Flare

CFHT SNLS

GALEX

Chandra(AEGIS)

Strongest empirical evidence for a stellar disruption event to date!

Gezari et al. 2006, in press

Constraints on MBH:MBH(t0-tD) > 1 x 108 Msun

MBH() ~ 1+1-.5 x 109 Msun

Critical upper limit on MBH for RT > Rs:Mcrit = 1 x 108 Msun (no spin)Mcrit = 8 x 108 Msun (O5 star)Mcrit = 8 x 108 Msun (with spin)

Upper limit on MBH for RBB > Rms:RBB < 4 x 1013 cmRms = 6Rg (no spin)Rms Rg (with spin)

If MBH > 108 Msun then RBB < 6Rg, and the black hole must have spin!

Disruption of a Star by a Disruption of a Star by a SpinningSpinning Black Hole Black Hole

2004.6 2004.9

NewNew Confirmed Flare from Inactive Galaxy Confirmed Flare from Inactive Galaxy

TOO VLT Spectrumcourtesy of Stephane Basa

Awaiting results from Chandra TOO!

z = 0.32

No Seyfert emission lines!

Dedicated Time Domain SurveyDedicated Time Domain Survey

• First time domain survey in UV• Designed to complement future ground-based TDS (PanStarrs, LSST)• All time-domain products, notably variable object alerts, will be immediately made public for community follow-up• Produce automated pipeline triggers to generate IAU and/or GCN circulars• Prevalence of UV-bright early evolution of flaring objects• The TDS may detect: supernovae, gamma-ray faint bursts, novae, macronovae, magnetic degenerate binaries, low mass x-ray binaries, chromospherically active stars, QSOs and AGNs, pulsating degenerates, luminous blue variables

Stay Tuned….Stay Tuned….

Candidate in D4Candidate in D42003 2005

2004.4 2006.0

Le Phare photo-z fits an elliptical galaxy at z=0.56

FUVflare=22.9

CFHT SNLScourtesy of Stephane Basa

Flare SED

Galaxy Host

• Chandra TOO X-ray observation in May 8, 2006 does not have a soft blackbody temperature.

• Follow-up spectrum of a ROSAT candidate also showed hard X-ray spectrum.

• Awaiting results from VLT optical spectrum.

T = 5 x 105 K

Image SubtractionImage Subtraction

Transient sources!

Method:1) Register images using a list of point source positions in both images.2) Subtract second image from reference image after polynomial spatial warping.3) Search difference image for sources above a threshold value with a correlation with the PSF of > 0.5.

Transient SourceTransient Source

Optically unresolved quasar!

CFHT SNLScourtesy of Stephane Basa

2005.0 2005.6

GALEX FUV band is sensitive to Rayleigh Jean’s tail of the soft X-ray blackbody emission.

A large K correction makes unextincted flare flux detectable by DIS out to high z.

6.5x10-4 yr-1 (MBH/106 M)-.25

(WM 2004)

E+S0 luminosity functionMBH = 8.1x10-5 (Lbulge/ L )0.18

(FS 1991, MT 1991, MF 2001)

Volume to which flares can be detected in a 10 ks DIS exposure

Yields 5 events yr-1 sq.deg-1(z ≤ 1)

Detection Rate with GALEXDetection Rate with GALEX

Transient SourceTransient Source

Optically resolved galaxy!

tmax≈2005.12

CFHT SNLScourtesy of Stephane Basa

2005.0 2005.6

SummarySummary

• Rich data set in spectral and temporal coverage when you combine GALEX + CFHT SNLS.

• Optical spectroscopy is necessary to confirm that flaring galaxies do not have an AGN, and we follow up our best candidates with Chandra TOO imaging.

• Legacy fields with multiwavelength coverage seem to be the most promising for identifying interesting variable sources.

• Can measure the distribution of masses and spins of dormant black holes.

• We have a new candidate with an inactive galaxy host confirmed by VLT TOO optical spectroscopy. We are awaiting the Chandra TOO observation results.