The variability of active galaxies: I. Broad-band noise X-ray power spectra from XMM-Newton and Swift

TL;DR

This study analyzes long-term soft X-ray variability in 56 unabsorbed AGN using XMM-Newton and Swift data and a Gaussian-process-based method called PIORAN to estimate bending power spectra. Bend frequencies span from roughly $7\ ext{min}$ to $62\ ext{days}$, with high-frequency slopes typically $\alpha_2>2$, and low-frequency bends detected in several sources, including new ones. The bend timescale scales approximately linearly with black hole mass $M_{ m BH}$ and shows only a weak dependence on luminosity, after accounting for mass, with intrinsic scatter indicating additional drivers or non-stationarity. The results align with MRI/GRMHD expectations for inner-disk variability and reveal parallels and distinctions with BH X-ray binaries, while highlighting the need for future UV/optical timing and more complex QPO modeling. Overall, the work provides a comprehensive, model-based reanalysis of AGN timing that enhances the X-ray variability plane and informs disk-physics interpretations.

Abstract

Accreting supermassive black holes at the centres of galaxies are the engine of active galactic nuclei (AGN). X-ray light curves of unabsorbed AGN show dramatic random variability on timescales ranging from seconds to years. The power spectrum of the fluctuations is usually well-modelled with a power law that decays as $1/f$ at low frequencies, and which bends to $1/f^{2-3}$ at high frequencies. The timescale associated with the bend correlates well with the mass of the black hole and may also correlate with bolometric luminosity in the `X-ray variability plane'. Because AGN light curves are usually irregularly sampled, the estimation of AGN power spectra is challenging. In a previous paper, we introduced a new method to estimate the parameters of bending power law power spectra from AGN light curves. We apply this method to a sample of 56 variable and unabsorbed AGN, observed with XMM-Newton and Swift in the $0.3-1.5$ keV band over the past two decades. We obtain estimates of the bends in 50 sources, which is the largest sample of X-ray bends in the soft band. We also find that the high-frequency power spectrum is often steeper than 2. We update the X-ray variability plane with new bend timescale measurements spanning from 7 min to 62 days. We report the detections of low-frequency bends in the power spectra of five AGN, three of which are previously unpublished: 1H 1934-063, Mkn 766 and Mkn 279.

Figures (10)

• Figure 1: Light curves and power spectral densities for Fairall 9, NGC 7469, NGC 4748 and 1H 1934-063. The sources are identified by a similar colour in all panels.
• Figure 2: Correlations between the black hole mass, optical luminosity, $2-6$ keV X-ray luminosity and the median value of the bend timescale $t_1$, low-frequency slope $\alpha_1$, high-frequency slope $\alpha_2$, power spectrum at the bend frequency, power spectrum at the bend frequency times bend frequency, integral of power spectrum model and average rms amplitude of variability $F_\mathrm{var}$ computed over segments of $30$ ks from XMM--Newton light curves. The distribution of each quantity are plotted in the upper and right panels.
• Figure 3: Posterior power spectra of the sources with evidence for two bends. The shaded areas correspond to the $68\%$ and $95\%$ percentiles from the posterior distributions. As we are mainly interested in the general shape and not total power, the amplitudes of individual sources have been rescaled for plotting purposes.
• Figure 4: Distribution of the power spectral parameters for sources with evidence of a second bend. $f_2/f_1$ is the ratio between the high- and low-frequency bends and $\alpha_i$ for $i\in[1,2,3]$ are the slopes of the bending power law model.
• Figure 5: X-ray variability plane for all sources. For sources with detection of low-frequency bend, we use the high-frequency bend.