The broad-band XMM-Newton and INTEGRAL spectra of bright type 1 Seyfert galaxies

TL;DR

This study analyzes broad-band X-ray spectra of nine bright type 1 Seyfert galaxies using XMM-Newton and INTEGRAL data, revealing typical AGN spectral features and constraining high-energy cut-offs.

Contribution

First simultaneous broad-band spectral analysis of these Seyfert galaxies combining XMM-Newton and INTEGRAL data, with detailed modeling of spectral components and high-energy cut-offs.

Findings

Most spectra are well described by a simple power-law model.

Soft components and FeK emission lines are common.

High-energy cut-offs are constrained below ~150 keV in four sources.

Abstract

The 0.5-150 keV broad-band spectra of a sample of nine bright type 1 Seyfert galaxies are analyzed here. These sources have been discovered/detected by INTEGRAL and subsequently observed with XMM--Newton for the first time with high sensitivity below 10 keV. The sample, although small, is representative of the population of type 1 AGN which are now being observed above 20 keV. The intrinsic continuum has been modeled using three different parameterizations: a power-law model, an exponential cut-off power-law and an exponential cut-off power-law with a Compton reflection component. In each model the presence of intrinsic absorption, a soft component and emission line reprocessing features has also been tested. A simple power-law model is a statistically good description of most of the spectra presented here; an FeK line, fully and/or partial covering absorption and a soft spectral component are detected in the majority of the sample sources. The average photon index (< Gamma > = 1.7 $\pm$ 0.2) is consistent, within errors, with the canonical spectral slope often observed in AGN although the photon index distribution peaks in our case at flat Gamma (~ 1.5) values. For four sources, we find a significantly improved fit when the power-law is exponentially cut-off at an energy which is constrained to be below ~ 150 keV. The Compton reflection parameter could be estimated in only two objects of the sample and in both cases is found to be R > 1.