Silicate emission in a type-2 quasar: JWST/MIRI constraints on torus geometry and radiative feedback

Type-2 quasars (QSO2s) are AGN seen through a significant amount of dust and gas that obscures the central supermassive black hole and the broad line region. Despite this, recent mid-infrared spectra of the central 0.5-1.1 kpc of five QSO2s at z~0.1, obtained with the MRS module of JWST/MIRI, revealed 9.7, 18, and 23 micron silicate features in emission in two of them. This indicates that the high angular resolution of JWST/MIRI now allows us to peer into their nuclear region, exposing some of the directly illuminated dusty clouds that produce silicate emission. To test this, we fitted the nuclear mid-infrared spectrum of the QSO2 with the strongest silicate features, J1010, with two different sets of torus models implemented in an updated version of the Bayesian tool {\tt BayesClumpy}. These are the CLUMPY and the CAT3D-WIND models. The CAT3D-WIND model is preferred by the observations based on the marginal likelihood and fit residuals, although the two torus models successfully reproduce the spectrum by means of intermediate covering factors ($\rm C_T=0.45\pm^{0.26}{0.18}$ and $\rm C_T=0.66\pm^{0.16}{0.17}$ for the CLUMPY and CAT3D-WIND models) and low inclinations ($\rm i=50^\circ\pm^{8^\circ}{9^\circ}$ and $\rm i=13^\circ\pm^{7^\circ}{6^\circ}$). Indeed, four of the five QSO2s with JWST/MIRI observations, including J1010, are in the blowout or ‘‘forbidden’’ region of the Eddington ratio-column density diagram, indicating that they are actively clearing gas and dust from their nuclear regions, leading to reduced covering factors. This is in contrast with Seyfert 2 galaxies observed with JWST, which are in the ‘‘permitted’’ regions of the diagram and show 9.7 micron silicate features in absorption. This supports a scenario where the more luminous the AGN and the higher their Eddington ratio, the lower the torus covering factor, driven by radiation pressure on dusty gas.

💡 Research Summary

This paper presents the first high‑resolution mid‑infrared (MIR) spectroscopic study of a type‑2 quasar (QSO2) that reveals silicate emission features directly from its nuclear region. Using the MRS mode of JWST’s Mid‑Infrared Instrument (MIRI), the authors obtained a 5–25 µm spectrum of the QSO2 J1010 (SDSS J101043.36+061201.4) with an angular resolution of 0.3–0.8″, corresponding to physical scales of 0.5–1.1 kpc at z ≈ 0.098. After carefully modeling and subtracting polycyclic aromatic hydrocarbon (PAH) bands and prominent emission lines, the residual continuum displays strong silicate emission at 9.7 µm, 18 µm, and a crystalline feature at 23 µm. The 9.7 µm silicate strength (S₉.₇ = 0.49 ± 0.01) is comparable to the strongest values seen in type‑1 AGN, indicating that hot dust (several hundred to ~1000 K) is directly visible in the nucleus despite the optical type‑2 classification.

To interpret these data, the authors employed an updated Bayesian fitting tool, BayesClumpy2, which implements Markov Chain Monte Carlo (MCMC) sampling to derive posterior probability distributions for torus model parameters. Two state‑of‑the‑art dusty torus frameworks were tested: (1) the classic CLUMPY model (Nenkova et al. 2008), which assumes a toroidal distribution of optically thick clouds, and (2) the CAT3D‑WIND model (Hönig & Kishimoto 2017), which adds a dusty wind component launched from the inner disc. Both models were fitted to the continuum spectrum, allowing for variations in cloud number, optical depth, radial distribution, torus angular width, and viewing angle, while CAT3D‑WIND also includes wind opening angle, wind cloud number, and wind‑to‑disc covering‑factor ratio.

Both models successfully reproduce the overall shape of the MIR spectrum, but the CAT3D‑WIND model yields a higher marginal likelihood and smaller residuals, especially around the 23 µm crystalline silicate band. The inferred torus covering factors are intermediate: C_T = 0.45⁺⁰·²⁶_₋₀·₁₈ for CLUMPY and C_T = 0.66⁺⁰·₁₆_₋₀·₁₇ for CAT3D‑WIND. The best‑fit viewing angles differ: CLUMPY prefers a moderately inclined line of sight (i ≈ 50° ± 8°), whereas CAT3D‑WIND favors a near‑face‑on view (i ≈ 13° ± 7°). These results imply that, even for an object classified as type‑2 in the optical, the MIR‑emitting dust is only partially obscuring the central engine, consistent with a clumpy or wind‑dominated geometry.

The authors place J1010 in the context of a broader sample of five QSO2s observed with JWST/MIRI. Four of these lie in the “blow‑out” or “forbidden” region of the Eddington‑ratio versus column‑density diagram, where radiation pressure on dusty gas is expected to expel material from the nuclear region. J1010 itself has a bolometric luminosity log L_bol ≈ 45.55 erg s⁻¹, a black‑hole mass log M_BH ≈ 8.4 M_⊙, an Eddington ratio λ_Edd ≈ 10⁻⁰·⁸, and a CO‑derived column density log N_CO ≈ 22.3 cm⁻². These values place it near the boundary between type 1.8–1.9 and type 2 AGN in terms of obscuration, supporting the idea that it is in a transitional phase where the torus is being cleared.

In contrast, a comparable JWST/MIRI study of six nearby Seyfert 2 galaxies (GA‑TOS sample) shows silicate absorption at 9.7 µm and 18 µm, deep ice and aliphatic grain features, and higher X‑ray column densities (log N_H ≈ 22.2–24.3 cm⁻²). These Seyferts occupy the “permitted” region of the Eddington‑ratio/column‑density diagram, indicating that radiation pressure is insufficient to significantly reduce their torus covering factors. The systematic difference between the QSO2s and Seyfert 2s thus provides observational evidence for a luminosity‑ and Eddington‑ratio‑driven evolution of the dusty torus: higher luminosities and λ_Edd lead to lower covering factors via radiative feedback.

The paper discusses limitations, notably the strong degeneracies among torus parameters (e.g., covering factor vs. inclination) and the restricted wavelength coverage (5–25 µm) that precludes simultaneous constraints from far‑infrared or X‑ray data. The wind parameters in CAT3D‑WIND are held fixed, limiting quantitative comparison with molecular outflows observed with ALMA. Future work should combine JWST MIR spectra with ALMA CO kinematics, X‑ray spectroscopy, and longer‑wavelength FIR photometry to break degeneracies and fully characterize the interplay between dusty winds, torus geometry, and AGN feedback.

In summary, the study demonstrates that JWST/MIRI can resolve silicate emission from the nuclear regions of type‑2 quasars, providing direct evidence that radiation pressure can thin the dusty torus and launch dusty winds in high‑luminosity AGN. The preference for the CAT3D‑WIND model underscores the importance of a wind component in shaping MIR spectra of luminous, high‑Eddington AGN, while the contrast with Seyfert 2 galaxies supports a unified evolutionary picture where torus covering factor declines as AGN power and Eddington ratio increase. This work thus marks a significant step forward in linking high‑resolution MIR observations to the physics of AGN feedback and torus evolution.