Variability from Xray binaries to AGN Phil Uttley

Variability: from X-ray binaries to AGN Phil Uttley University of Southampton With thanks to: Ian Mc. Hardy, Patricia Arévalo

Overview • The remarkable connection between BHXRB and AGN X-ray variability • Caveats and surprises (AGN are different in at least one respect…) • Interpreting AGN continuum spectral variability: evidence for multiple variability time -scales,

X-ray variability in accreting black hole systems X-ray light curve of NGC 5506 X-ray light curve of Cyg X-1

Black hole XRB spectral states (Done & Gierlinski 2007)

Quantifying variability: the ‘power spectrum’ (PSD) of Cyg X-1 Soft state Hard state

A gallery of black hole XRB PSDs Done & Gierlinski 2005

The origin of the variability Bottom line: it’s fluctuations in the accretion flow (disk or ADAF/accreting corona) - please drop the ‘flaring solar corona’ analogy! • Very broad range of variability time-scales • Neutron-star/BH timing similarities (e. g. Klein-Wolt & van der Klis 2007) • Rms-flux relations/non-linearity (e. g. Uttley et al. 2005) • Optical/X-ray variability in AGN • ‘Flickering’ is standard outcome of MHD simulations • Spectral-timing properties…

Building the PSD one disk-annulus at a time

Ark 564

RXTE 1995 -2009?

Characteristic time-scales scale with black hole mass >108 M 4 107 M 106 M

Characteristic time-scales scale inversely with accretion rate Markowitz & Uttley 2005 For AGN with the same BH mass, PSD break time -scale appears to depend on luminosity Uttley & Mc. Hardy 2005 Time-scale normalised by BH mass scales inversely with Lbol/LEdd

The BH variability fundamental plane: mass-luminosity-timescale Mc. Hardy et al. , Nature, 2006 BH XRBs Yes! t. B MBH 2/Lbol Now with more detailed fitting: Does break time -scale t. B scale with bolometric luminosity’ of AGN? Try: t. B MBHA/Lbol. B AGN A B

Variability vs. Luminosity Black squares: NLS 1, Crosses: BL Seyferts (Turner et al. 1999) 2 rms power frequency observed band Log L 2 -10 ke. V frequency Can use ‘excess-variance’ to estimate mass [e. g. O’Neill et al. , Gierlinski et al. 2007, Nikolajuk (this conf. )] but should take accretion rate into account!

Power covers a broad bandwidth: soft-state like fake intermediate

Problems & caveats • • Is Cyg X-1 unusual, is it really ‘soft state’? Other soft state BHXRBs show much lower rms (few per cent or less), c. f. AGN 20 -50% rms Are Seyferts really soft state? Maybe normal Seyferts hard state, narrow line Seyfert 1 s soft? (e. g. Gierlinski et al. 2007) GRO J 1655 -40 X-ray Flux Hardness Fractional rms

Are soft states variable enough? Cyg X-1 disk+PL energy spectra mean Cyg X-1 PSDs 6 -13 ke. V 2 -6 ke. V rms The variability is produced primarily by the power-law, not the disk, which is ~constant & dilutes variability - don’t expect in AGN (disk in optical/UV, but see later…)

Are soft states variable enough? Cyg X-1 Variability even more heavily diluted by very large constant disk in other soft-state sources GRO J 1655 -40 (Done et al. 2007)

How typical is the Cyg X-1 soft-state PSD?

Are Seyferts in soft states? NGC 5548 (Gaskell et al. 2008) Yes, most/all have ‘soft’ SEDs (mdot> 1%) Low/hard states are LINERs/FR Is?

Surprise: AGN disks are variable! Not X-ray reprocessing! Simple energetics argument: Long-term optical fractional variability amplitude > X-ray, absolute luminosity variations even larger! (NGC 5548, From Uttley et al. 2003) But 1 year = few seconds in BHXRBs…not observed. AGN soft state disks unstable, BHXRB stable?

What about reprocessing? 80% of incident flux 0. 5% of incident flux 20% of incident flux Disk reflection implies a thermal reprocessed component

Optical still requires some reprocessing… (phew!) MR 2251 -178 (Arévalo et al. , submitted): good X-ray/opt correlation but small lags of short-term variations require small (<30%) reprocessed variability component

Understanding complex spectral variability in AGN Example: Ark 564 shows complex spectral variability and time lags (Dewangan 2007)

Cyg X-1 hard state spectral-timing behaviour PSD Basic observations: Multiple-Lorentzian Power spectrum (PSD) Harder X-rays more power at high frequencies (short time -scales) Ratio of soft/hard PSDs Hard lags soft, lag is frequency dependent 8 -13 ke. V vs 2 -4 ke. V lags Spectral-timing shifts between Lorentzians

PSD energy dependence in Ark 564 (Mc. Hardy et al. 2007) Soft PSD (0. 6 -2 ke. V) Hard PSD (2 -10 ke. V) PSD is ‘bandlimited’ (unlike soft state). Can be fitted by two Lorentzians, higher frequency component stronger at higher energies

Complex X-ray lags in Ark 564 (Mc. Hardy et al. 2007) Lag (s) In Ark 564, lags seem to show similar ‘stepped’ behaviour to Cyg X-1, this time may be ‘very high’ state (also multi. Lorentzians) Frequency (Hz)

Mapping the emission region Time * Travel time delay from origin of signal Flux emitted per unit travel time = ? flux mdot ‘Response Function’ Time The lag between two bands is the difference between emissivity-weighted propagation times (Arévalo & Uttley 2006), which is a function of radius of origin of the signal and hence the signal time-scale.

Conclusions • AGN/BHXRB variability is consistent with being the same type of process, scaled by mass and mdot • Seyfert variability and SEDs are consistent with soft states • AGN disks are intrinsically unstable on short timescales, unlike in soft state BHXRBs • Many properties of AGN variability may already be known from BHXRBs, which will help to decode the variability