Infrared Signatures of Dusty Tori Formed by MHD-Driven Outflows

TL;DR

This study uses JWST mid-infrared spectra and a new radiative transfer model to explore the structure of dusty tori in AGN, revealing diverse density profiles and their impact on AGN luminosity estimates.

Contribution

It introduces a physically motivated 3D radiative transfer framework to test MHD wind scenarios with different radial density laws, uncovering structural diversity in AGN tori.

Findings

Majority favor flat density profiles ($p=0.5$)

Significant minority prefer steep profiles ($p=2.0$)

Radial profile assumptions greatly affect AGN luminosity estimates

Abstract

We investigate the radial structure of active galactic nuclei (AGN) dusty tori by forward-modeling mid-infrared (MIR) spectra observed with JWST/MIRI-MRS for a sample of 25 AGN. We develop a physically motivated three dimensional radiative transfer framework to test various magnetohydrodynamic (MHD) wind scenarios characterized by radial density laws of the form $n(r)\propto r^{-p}$ for $p \in \{0.5, 1.0, 1.5, 2.0\}$. Our Bayesian analysis reveals a pronounced structural dichotomy rather than a single universal density law: a majority (14/25) of the sources favor the flattest profile ($p=0.5$), while a significant minority (8/25) statistically prefers the steepest, most compact distribution ($p=2.0$). We also identify a systematic redward shift in the silicate emission peak, consistent with the presence of micron-sized, processed grains in the inner regions of the wind. Critically, we find that the inferred AGN fractional contribution, $f_{\rm AGN}$, is highly sensitive to the assumed radial index, exhibiting up to a four-fold systematic shift between $p=0.5$ and $p=2.0$. This p-degeneracy highlights that static structural assumptions can significantly bias AGN power estimates. The identified structural diversity suggests that MHD-driven winds may exist in multiple dynamical states, though the precise physical drivers governing these configurations remain to be uniquely determined. These results underscore the limitations of single-profile models and emphasize the need for high-resolution spatial constraints to break these fundamental degeneracies.