Relativistic X-ray reverberation modelling of the combined time-averaged and lag-energy spectra in AGN

TL;DR

This paper presents relativistic ray tracing simulations to model X-ray reverberation in AGN, fitting both spectra and lag-energy data simultaneously to better understand black hole environments.

Contribution

It introduces a combined spectral and timing fitting approach using relativistic models, including ionization gradients, for the first time in AGN reverberation studies.

Findings

Ionization gradients explain lag-energy features with 3 keV and 7-10 keV dips.

Model fits suggest source heights > 5 gravitational radii or highly ionized inner disks.

Simultaneous spectral and timing fitting yields more consistent AGN parameters.

Abstract

General relativistic ray tracing simulations of the time-averaged spectrum and energy-dependent time delays in AGN are presented. We model the lamp-post geometry in which the accreting gas is illuminated by an X-ray source located on the rotation axis of the black hole. The spectroscopic features imprinted in the reflection component are modelled using REFLIONX. The associated time delays after the direct continuum, known as reverberation lags, are computed including the full effects of dilution and ionization gradients on the disc. We perform, for the first time, simultaneous fitting of the time-averaged and lag-energy spectra in three AGN: Mrk 335, IRAS 13224-3809 and Ark 564 observed with XMM-Newton. The best fitting source height and central mass of each AGN partly agree with those previously reported. We find that including the ionization gradient in the model naturally explains lag-energy observations in which the 3 keV and 7-10 keV bands precede other bands. To obtain the clear 3 keV and 7-10 keV dips in the lag-energy profile, the model requires either a source height > 5$r_g$, or a disc that is highly ionized at small radii and is colder further out. We also show that fitting the lag or the mean spectra alone can lead to different results and interpretations. This is therefore important to combine the spectral and timing data in order to find the plausible but self-consistent fits which is achievable with our model.