The Hard X-Ray Luminosity Function of High-Redshift (z > 3) AGN Fabio Vito Dipartimento di Fisica e Astronomia - Università di Bologna INAF-OABo F Vito, RGilli, C Vignali, A Comastri, M Brusa, N Cappelluti, K Iwasawa; MNRAS accepted yesterday! (on astro-ph today)
Introduction AGN Population Evolution z < 3: AGN downsizing (LDDE; eg Miyaji+00, Ueda+03, Hasinger+05, La Franca+05, Silverman+08, Ebrero+09, Yencho+09, Ueda+14) z > 3: not well assessed Ueda+14
Introduction High-Redshift AGN Population Evolution Need for large, complete and reliable samples Deep (4 Ms CDF-S) and wide (XMM-COSMOS, Chandra-COSMOS, SXDS) X-ray surveys Multi-wavelength coverage
3 z < 51 AGN sample AGN sample 141 X-ray (05 2 kev) detected AGN at 3 z < 51 5 x Hiroi+12 sample 2 x Civano+11 sample using the most up-to-date redshift information and a careful and clean (eg no strong radio-loud AGN) selection in the 4 Ms CDFS (Vito+13), XMM-COSMOS (Brusa+10, Salvato+11, Civano+12), Chandra-COSMOS (Civano+12, Lilly+ in prep) and SXDS (Hiroi+12) fields Redshift completeness > 95%
3 z < 51 AGN sample Spectral parameters Γ = 18 fixed N H from uniform spectral analysis (CDFS, C-COSMOS) or HR (SXDS, Hiroi+12) Objects from X-COSMOS assumed unobscured Intrinsic L 2 10 kev from spectral analysis (CDFS, C-COSMOS), literature (SXDS, Hiroi+12), or extrapolated from soft-band flux from catalogue (X-COSMOS, Cappelluti+09)
The Hard X-Ray Luminosity Function φ = φ U + φ A φ as in Page&Carrera 2000 logn H < 23 UNABSORBED (105 objects) Ω U logn H > 23 ABSORBED (36 objects) Ω A
The Hard X-Ray Luminosity Function Evolutionary models Φ(L, z) = [( L A(z) ) γ1 ( + L ) γ2 ] L (z) L (z) PLE : L (z) L (3) (1 + z) p lum PDE : A(z) A(3) (1 + z) p den ILDE : PLE + PDE LADE : PLE + A(z) A(3) 10 p den (1+z) LDDE : A(z) A(3) (1 + z) p den +β(logl 44)
The Hard X-Ray Luminosity Function Fit to unbinned data through a Maximum Likelihood procedure PLE PDE ILDE LADE LDDE MODEL 2DKS PLE 005 PDE 038 ILDE 038 LADE 046 LDDE 042 2DKS 020 model and data not significantly different
The Hard X-Ray Luminosity Function Correction for redshift incompleteness: Θ = Θ(z, L X, N H ) 65 sources with no redshift information (but flux known) All of them assumed to be at 3 < z < 51 Assumed same fraction of absorbed sources (at similar fluxes) and same redshift distribution of the sources with redshift BEFORE CORRECTION AFTER CORRECTION BEFORE CORRECTION AFTER CORRECTION
The Hard X-Ray Luminosity Function Fit to unbinned data after completeness correction PLE PDE ILDE LADE LDDE MODEL 2DKS PLE 020 PDE 044 ILDE 020 LADE 023 LDDE 027
Obscured AGN fraction F (lognh >23) = φ A /(φ A + φ U ) Constant with luminosity (χ 2 fit returns F 23 = 054 ± 005) but the most obscured sources at low-l are probably missing! Evolution from z = 0 to z > 3
Space density Φ = dn dv = N i=1 1 V max,i Φ (1 + z) p with p = 600 +084 087, in agreement with Hiroi+12, but larger dataset and different method (Maximum Likelihood fit on unbinned data vs χ 2 minimization) We confirm the decline in the space density of luminous high-redshift AGN (factor of 10 from z=3 to 5)
Space density No evidence for the up-sizing (ie flatter space density for less luminous AGN at z > 3) suggested by Ueda+14 Larger sample of L X < 10 44 AGN needed to discriminate between PDE and LDDE
Conclusions Conclusions Evolution of the HXLF at z > 3 dominated by a negative density term (PDE) More complex models mimic the behaviour of the PDE model Larger samples of low-luminosity (L X < 10 44 ) AGN needed to constrain a possible luminosity-dependent density evolution at high-z Obscured (logn H > 23) AGN fraction is 054 ± 005 No evidence for an anti-correlation with luminosity, but evolution from z = 0! The space density of luminous AGN declines by a factor of 10 from z=3 to 5 No evident dependency of the decline slope on luminosity, but larger low-l samples are required Future perspective Larger samples of (low-luminosity) AGN thanks to the 7 Ms in CDFS (+ deep follow-up and/or proper photo-z)
Conclusions Fit parameters MOD A L γ 1 γ 2 p lum p den β 2DKS PLE 065 +006 006 656 +238 222 021 +016 020 258 +075 060 373 +077 092 005 PDE 110 +011 011 526 +106 120 022 +013 016 379 +108 087 600 +084 087 038 ILDE 113 +011 011 513 +132 153 021 +013 017 375 +110 091 013 +094 081 613 +117 123 038 LADE 108 +011 011 510 +133 154 021 +013 018 374 +111 091 016 +095 082 057 +011 011 046 LDDE 105 +010 010 524 +105 119 028 +016 019 387 +108 088 643 +112 117 118 200 +206 042
Conclusions Binned HXLF φ = Φ dlogl = d2 N dv dlogl Assuming φ does not vary in the z LogL bin (ie narrow z LogL bin) φ ( z, LogL) = N( z, LogL) V ( z, LogL) = N( z, LogL) ΩΘ dv dz dz dlogl logl z
Conclusions Binned HXLF: Page&Carrera2000 vs V MAX (Page&Carrera, 2000)
Conclusions Likelihood estimator L = 2 N i=1 ln[φ(z i, L i )] + 2 φ(z, L)ΩΘ dv dz dzdl (Marshall+1983)