The remarkable X-ray variability of IRAS 13224–3809 – I. The variability process

Abstract

We present a detailed X-ray timing analysis of the highly variable NLS1 galaxy, IRAS 13224-3809. The source was recently monitored for 1.5 Ms with XMM-Newton which, combined with 500 ks archival data, makes this the best studied NLS1 galaxy in X-rays to date. We apply standard time-and Fourier-domain techniques in order to understand the underlying variability process. The source flux is not distributed lognormally, as expected for all types of accreting sources. The first non-linear rms-flux relation for any accreting source in any waveband is found, with rms ∝ flux 2/3 . The light curves exhibit significant strong non-stationarity, in addition to that caused by the rms-flux relation, and are fractionally more variable at lower source flux. The power spectrum is estimated down to ∼ 10 −7 Hz and consists of multiple peaked components: a lowfrequency break at ∼ 10 −5 Hz, with slope α < 1 down to low frequencies; an additional component breaking at ∼ 10 −3 Hz. Using the high-frequency break we estimate the black hole mass M BH = [0.5 − 2] × 10 6 M , and mass accretion rate in Eddington units, m Edd ∼ > 1. The broadband PSD and accretion rate make IRAS 13224-3809 a likely analogue of Very-high/Intermediate state black hole X-ray binaries. The non-stationarity is manifest in the PSD with the normalisation of the peaked components increasing with decreasing source flux, as well as the low-frequency peak moving to higher frequencies. We also detect a narrow coherent feature in the soft band PSD at 7×10 −4 Hz, modelled with a Lorentzian the feature has Q ∼ 8 and an rms ∼ 3 %. We discuss the implication of these results for accretion of matter onto black holes.

A dynamic black hole corona in an active galaxy through X-ray reverberation mapping

“…The background was extracted from a large rectangular region on the same chip, placed away from the source and avoiding the Cu ring on the EPIC-pn chip. We produce background-subtracted light curves in the 0.3-1.0 keV (soft) and 1.2-5.0 keV (hard) bands (see 7 for more details). Details of the broadband spectral analysis can be found in 11,16 .…”

“…Whilst these models are able to reproduce some aspects of the spectral variability, they fail to describe the fundamental properties of the variability process observed in accreting systems: the lognormal distribution of fluxes and the linear rms-flux relation 7,13,[52][53][54] . The rms-flux relation is a universal signature of an accretion disc and is observed in many wavebands and in all types of accreting compact objects, such as non-jetted AGN, but also including accreting white dwarfs, young stellar objects, as well as neutron star and black hole X-ray binaries 7,13,[55][56][57][58][59][60][61][62][63] . Magneto-hydrodynamic (MHD) accretion disc simulations are able to reproduce the rms-flux relation 64,65 .…”

“…However, there is an inherent degeneracy in the reverberation signal between black hole mass, inner disc radius and height of the illuminating corona above the disc. Here, we use a long X-ray observation of the highlyvariable active galaxy, IRAS 13224-3809, to track the reverberation signal as the system evolves on timescales of a day 7,8 . With the inclusion of all the relativistic effects, modelling reveals that the height of the X-ray corona increases with increasing luminosity, providing a dynamic view of the inner accretion region.…”

“…IRAS 13224-3809 has been observed for 16 full orbits (~130 ks per orbit) with the European Space Agency's X-ray Multi-Mirror Mission (XMM-Newton 15 ), totalling 2 Mega-seconds of observations. The source is one of the most variable X-ray objects in the sky, undergoing rapid and large amplitude variations on timescales of minutes 7,8 . It is unique in that the shape of the variability (visualised as power spectral density -PSD) is also time-dependent, varying on days timescales 7,8 .…”

Deviations from normal distributions in artificial and real time series: a false positive prescription

“…The interpretation of this is that the steepening PSD spectral index progressively diminishes the effect of high frequency components, effectively reducing the total number of degrees of freedom in each simulated time series (see also Uttley et al 2005). More recent work by Alston (2019) has shown that the exponential flux distributions at steep PSD indices can deviate from the expected lognormal distribution. Analytically, this can be explained by first obtaining the total power in a time series by integrating the PSD (Eqn.…”