The BAT AGN Survey - Progress Report J. Tueller, C. Markwardt, L. Winter and R. Mushotzky Goddard Space Flight Center The Swift BAT (Burst and Transient Telescope) has been observing the whole sky in the 15-200 kev band for ~22 months With follow-up x-ray and optical and IR observations- this is a progress report The first unbiased survey of AGN in the local universe- no selection effects due to obscuration, galaxy properties or optical or radio properties. These data allow a direct comparison of selection effects for AGN across the electromagnetic spectrum since the majority of the objects are close and bright Large (~ 500 by end of next year) all sky unbiased sample of low redshift AGN Blazars over wide z range Uniform selection criteria Objects are bright and easily studied in all wavelength bands Rare objects (e.g. type II QSOs, very high z Blazars) Flux limit ~1-3x10-11 ergs/cm 2 /sec 15-200 kev (~1 mc)
The Local Census of Active Galaxies-aka Radiating Massive Black Holes J. Tueller, C. Markwardt- RFM and L. Winter The change in the luminosity and number of AGN with time are fundamental to understanding the origin and nature of massive black holes and the creation and evolution of galaxies ~20% of all energy radiated over the life of the universe comes from AGN- a strong influence on the formation of all structure. A large fraction of all the AGN and their radiation comes from objects which are obscured from view in the optical/uv X-ray Color Image (1deg) of the Chandra Large Area X-ray Survey- CLASXS-400ks, 525 sources
Why is the Swift/BAT census of Black Holes desirable? Hard X-rays are a unique signature of accreting black holes Wide field finds rare objects - type II QSOs hard X-rays unaffected by absorption yielding a complete census hard X-ray Image The last all-sky hard X-ray survey was HEAO1 in 1977 BAT is 17 times more sensitive. detect rare sources high galactic latitudes for optical follow-up UV image complete x-ray follow-up with Swift/XRT and UVOT X-ray Image XRT joint spectra BAT
X-ray Selection of Active galaxies X-ray and optical image of a nearby AGN NGC4051- very high contrast in the x-ray image the upper limit of x-ray luminosity of ULXs ~5x10 41 ergs/sec and of entire starburst galaxies ~3x10 42 ergs/sec All nuclear sources with L(x > 10 42 are AGN Right now we know more about x-ray selected AGN at z~0.8 than at z~0 L X vs. redshift Spectroscopic Desert I=15-20 I=20-22 I=22-23 I>23 Brandt & Hasinger. 2005
Black Hole Finder The absorbing material can have very large column densities block soft x-rays and UV/optical making sources optically invisible. Chandra data show that there are >7x more hard x-ray selected than optically selected AGN (at same optical threshold) The most numerous AGN (L x <10 44 ergs/sec) evolve inversely from the well studied quasars and are more numerous in the local than high z universe Log N H =24.25 Log N H =24.75 Log N H =25.25 Wilman & Fabian (1999)
The Dark Side of AGN Many AGN are obscured- obscuring material is of several types Located in the ISM of the host galaxy A wind associated with the AGN Perhaps a obscuring torus Etc Lack of uniform sample not sensitive to absorption or emission from this structure has limited knowledge physical conditions in obscuring regions are not the same from object to object - can be complex with large and unpredictable effects on the spectrum
Nature of Hard X-ray selected sources Followed up Swift BAT selected sources with XMM, Suzaku and XRT Wide range of x-ray spectra Many of the IDs have no optical evidence for activity in literature even though they are very low z bright galaxies No correlation with Rosat flux XMM + BAT spectra Obvious why soft and hard x- ray band are uncorrelated
Why do We Need XMM Follow-Ups All other AGN surveys are biased (!) only by following up the BAT and Integral sources can one obtain the true distribution of x-ray spectral properties necessary for solving all the AGN science problems Only XMM and Suzaku have sufficient collecting area to obtain good quality spectra in short (<40ks) exposures (with loss of XIS 2 XMM has ~30% more collecting area than Suzaku) Only XMM can do lots of ~10ks exposures which are sufficient for these bright sources to characterize the.3-10 kev spectrum Only XMM has simultaneous optical/uv data necessary for the SED and estimate of star formation rate. Better spatial resolution of XMM than Suzaku can be important Suzaku is important for characterization of reflection and high energy spectrum simultaneous with x-rays
9-month Swift/BAT Survey 1-1.5 Ms > 1.5 Ms Ecliptic Plane 0.5-1 Ms Exposure Map Covers whole sky, mostly >1Ms deficit on Ecliptic Plane due to Sun avoidance Sensitivity improves as square root of time (1.2-2 X statistical) to 0.6 millicrab in 3 years Sensitivity vs Exposure Time 8.5 mcrab (T/20 ks) -0.5
Future of AGN Surveys Do we have a sufficient understanding of the z=0 AGN universe to understand the Spitzer/Herschel/JWST/ALMA sources- how does the IR connect to the rest of the spectrum NGC 3079 N4388 type II Sloan and Spitzer images understand the Nustar/Simbol-X sources. High signal to noise high spatial resolution only possible at low z with bright sources Provide large list of targets for Herschel, NeXT, Con-X and XEUS Need bright interesting sources for high signal to noise high spectral resolution studies Provide list of targets for IFU studies with new large ground based telescopes ESO 323-077 type-i NGC788 type II Spitzer spectra 5-40µ
How do the sources differ? The spectral energy distribution (SED) of the broad line and absorbed sources are very different This also changes with x-ray luminosity - lower luminosity sources are redder Absorbed objects objects contribute ~2/3 of today s rest mass of black holes ISO absorbed SCUBA
What are the optical properties of the X-ray AGN? HST GOODs survey allows first estimate of nuclear optical fluxes (from the ground the objects are too blurry) For the broad line objects (squares) x-ray and UV flux are strongly correlated For the narrow line objects optical and x-ray fluxes are uncorrelated and many of the optical nuclei at z~0.7 are invisible with deep HST observations. Relative nucleus vs galaxy Negative values mean nuclear dominated 126/286 Nuclei Not detected by HST in z Nucleus is often >5 mags fainter than expected at 3000A (rest frame) if due to extinction N(H)~10 22 atm/cm 2 for MW dust to gas ratio - very hard to study at high z- need low z objects
9 months of data --------------------- 323 sources 39 unidentified sources 158 galactic 5 galaxy clusters 158 AGN 16 beamed BAT Source Detections Galactic b >15 (74%) ---------------------- 159 sources 1 unidentified source 29 galactic 2 galaxy clusters 121 AGN 15 beamed 2 years of survey will be available soon. astro-ph 3rd Integral catalog with 450 sources all sky with 3.5 yrs of data 17 sources before BAT and Integral
Galactic Center Image (40x40 deg 2 )
Galactic Center Region
BAT Survey Characteristics Survey is continuing- now at 22 month 4 energy bands now expanded to 8 bands extragalactic BAT sources are easy to identify at high latitudes with x-ray positions bright IR galaxy low redshift (z<0.06) optical AGN or high absorption Exception - Blazars which can be faint (`19th mag) XRT positions and joint XRT/BAT spectra make ID's unambiguous XRT joint spectra BAT OM XRT BAT
threshold flux is ~2-3 x 10-11 erg cm -2 s -1 no obvious bias in index or N(H) with flux BAT bias if any is for harder sources No BAT selection of Index or N(H) with Flux Log F(X) Compton thick Log N(H) BAT AGN XMM Follow-up
Median redshift of identified sources is ~0.025- this is a very low z sample Sy1's have greater redshifts than Sy2's thus higher luminosity in the 20-100 kev band Redshift Distribution XMM + BAT spectra
Redshift of BAT Blazars 17 BL Lac, QSO, and blazars (all-sky) Blazar redshift distribution very different from Seyfert population 6 high redshift blazars detected (z >2), 4 not previously identified Tend to be optically dim (m~19th mag) 22 month source z=3.86 z Seyfert ~0.03 Number of objects Seyferts
Why BAT for AGN BAT data select AGN solely on the basis of their hard x- ray emission irrespective of their other properties Optical spectrum of NGC4992- A BAT source To confirm the correctness of the BAT identifications have followed up with Swift and XMM x-ray observations program of optical observations to understand the nature of these objects Objects tend to be IR/optically bright (except Bl Lacs)- easy to study at other bands Also easy to get images
SWIFT BAT Survey Compared to Other X-ray Surveys To first order the x-ray to optical ratio of the BAT sources is consistent with that seen in deep x-ray surveys BAT sources tend to be optically bright- SDSS +6dF spectra Brandt & Hasinger 05 Log(F X /F I ) +1 0-1 -2 Sy1 Sy2 I-mag calculated from 2MASS K-mag assuming K-I = 2 (Ferraras et al. 99)
BAT GALAXY Identifications so far (** not complete) Definition of absorbed Log N H > 22 Gal b >15 Sy1 Sy1.5 Sy2 no Sy ID ** abs 3 5 36 9 unabs 31 11 3 4 total 34 16 40 13 total 53 49 103 70% of the galaxies with no optical Sy classification are absorbed. Most unclassified galaxies are absorbed objects or edge on spiral galaxies with no lines (several with no spectroscopy) Sy2/Sy1 ratio is 1.2 (0.5 Sy1.5/Sy1) and 52% of Seyfert galaxies are absorbed confirming Markwardt et al 2005 Much lower than 75% predicted by the standard model for the CXB and unified models (Treister and Urry 2005)
Unabsorbed n H = 1023 cm -2 Fe K eqw = 450 ev Mkn 352 ESO 362-G018 Compton thick Fe K eqw = 260 ev NGC 1142 CGCG 041-020 n H = 1023 cm -2 XMM follow-up: -Range of fluxes (F X (0.3-10 kev)) from 1.6x10-12 to 3.0x10-11 erg/s cm 2-8 unabsorbed sources (n H < 10 21 ) -14 absorbed sources, 5 of which are Compton thick -Similar data exists for ~80 other sources
XMM has ~15x the counting rate of the Swift XRT data It is virtually impossible to detect Fe K lines or spectral complexity with the XRT data alone. Comparison of Quality of Data The possibly Compton-thick source Mrk 417 has a lower than expected EW for Fe K-α. PN (black), XRT (red), and BAT (green) spectra are fit with a partial covering absorption model (n H = 9x10 23 cm -2, Covering Fraction = 0.995, Γ = 2.3, along with gaussians at E = 6.41 and 6.74 kev). 6300 s XRT spectrum has too few counts (shown with 20 cts/bin) to characterize the Fe K- α line
Comparison of K band and BAT Luminosities The K band light consists of 2 components AGN light Starlight converting from starlight to mass (Novak et al 2006) the BH mass is M BH ~ 10 9 at M K ~ -25 where L(BAT)~43.5 Type II QSOs Using a bolometric correction of 25 gives L/L Edd ~0.02 There is a set of objects of very low L(BAT)/M(K)- both very low Eddington ratio - and Compton thick objects with suppressed hard x-rays. All but one of the high L(BAT ) sources are radio loud L(K) stars
The Distribution of Obscuration Unified models of active galaxies predict the relative number of different types of AGN (so-called types I and II) the ratio of obscured to un-obscured objects in the local universe should be 4:1 (Antonucci and Miller 1989, Treister et al 2006) BAT finds 1.2:2 - a serious discrepancy requiring modification of the unified modelneed good quality x-ray data to properly model N(H) N(H)=10 22 = ~4 A v Number of objects NH Distribution for Swift/BAT AGN unabsorbed (49) absorbed (52) (15) optically thick?
Heavily Absorbed and Complex Spectra Abound >25% of BAT sources have soft X- ray emission-spectral models used in XRB synthesis are wrong only a survey at >20 kev is unbiased by absorption Ph/cm2/sec/kev XMM BAT peaks at 20 kev 1 10 100 Energy (kev) Many have extremely complex spectra with soft and hard components - seem unrelated- need high quality x- ray spectra- low S/N data are highly misleading
Tests of the Standard Model With <E>~50 kev BAT measures the true nature of the continuum relatively unaffected by absorption or scattering BAT selected Sy1's have softer spectra than Sy2's (5.7σ) BAT selected Sy1's have higher luminosity than Sy2's (3.6σ) no selection effect for BAT
Tests of Unified Model The Unified model predicts that the obscuration is geometric -no dependencies of the obscuration on the luminosity of the AGN BAT lower luminosity sources are more likely to be obscured than the high luminosity sources Consistent with XMM and Chandra results on fainter higher z sources Fraction of objects
All the high N(H) high luminosity sources but one are radio loud e.g. Cyg-A Relation of Luminosity and N(H)
The unified model predicts always lines of sight in the torus which allow light escape without absorption This produces the soft x-ray components from scattering and the light to produce [OIII] + other lines So far in the BAT survey we have ~6 objects which have no soft componentsthe absorber is 4π! This objects are thus completely enshrouded (Ueda et al 2007) New Type of AGN? Swift J0601-8636 Swift J0138
Degeneracy of Model Fits Broad band fits to XMM + BAT data- no unique fits for many sources
Unexpected high redshift AGN! XMM Slew Survey Sources- K. Lewis z=0.39, L=4x10 45 erg/s No AGN lines z=0.76, L=10 46 erg/s No soft X-rays!
BAT Spectra Softer than 2-10 kev X-ray (BAT biased to harder spectra) BAT power law index consistently softer than the 2-10 kev index (RXTE simple Power law fits) median x-ray=1.74 median BAT=1.96 As predicted by reflection models- x-ray spectrum Σ of Pl + reflection, reflection less important at E> 40 kev so see true continuum form Equal slope Break=Γ bat -Γ xray
LogN/LogS and Luminosity Function ~800 sources >10-11 errors 25% in normalization, ~10% in slopes and <1% in break luminosity New, much tighter constraints test CXB models- in particular ratio of abs:unabs sources Two models that predict the XRB make different predictions for source counts at L BAT >10-11 ergs cm -2 s -1 Treister, Urry, and Lira: standard unified AGN model predict 2500 AGN Ghandi model predict 800 AGN BAT measures 1100 AGN
So How Much Time 500 sources - 10ks each for discovery quality spectra 5 Ms Why 500?? Wide range in flux and perhaps redshift Follow up spectra at 40ks each for selected subsample of complex spectra (1/3 of sample) 6 Ms Sample will be finished in 3 yearswhile do not know where systematics limit enter. So ~ 11 Ms to complete the project Very desireable to have Suzaku data for some of these to constrain Compton scattering/thickness