Quasar Accretion Disks Are Strongly Inhomogeneous

TL;DR

This paper presents inhomogeneous accretion disk models with large local temperature fluctuations that explain observed quasar variability, size discrepancies, and UV emission without requiring additional components.

Contribution

It introduces toy models of quasar disks with independent, large-amplitude temperature fluctuations that reconcile multiple observational discrepancies.

Findings

Inhomogeneous models match observed quasar variability and size estimates.

Spectral fits to quasar data are improved without invoking Compton atmospheres.

Microlensing simulations are consistent with observed light curves and predict detectable deviations.

Abstract

Active galactic nuclei (AGN) have been observed to vary stochastically with 10-20 rms amplitudes over a range of optical wavelengths where the emission arises in an accretion disk. Since the accretion disk is unlikely to vary coherently, local fluctuations may be significantly larger than the global rms variability. We investigate toy models of quasar accretion disks consisting of a number of regions, n, whose temperatures vary independently with an amplitude of \sigma_T in dex. Models with large fluctuations (\sigma_T=0.35-0.50) in 100-1000 independently fluctuating zones for every factor of two in radius can explain the observed discrepancy between thin accretion disk sizes inferred from microlensing events and optical luminosity while matching the observed optical variability. For the same range of \sigma_T, inhomogeneous disk spectra provide excellent fits to the HST quasar composite without invoking global Compton scattering atmospheres to explain the high levels of observed UV emission. Simulated microlensing light curves for the Einstein cross from our time-varying toy models are well fit using a time-steady power-law temperature disk, and produce magnification light curves that are consistent with current microlensing observations. Deviations due to the inhomogeneous, time-dependent disk structure should occur above the 1% level in the light curves, detectable in future microlensing observations with millimag sensitivity.