AGN STORM 2. XI. Spectroscopic reverberation mapping of the hot dust in Mrk 817

TL;DR

This paper presents the first intensive spectroscopic near-IR reverberation mapping of the hot dust in Mrk 817, using 157 cross-dispersed spectra across three telescopes to separate disk and dust components. It finds a dust reverberation radius of $R_{\\rm dust} \\sim 90$ light-days and a mean dust temperature of $T \\approx 1395$ K, consistent with carbonaceous dust and a luminosity-invariant inner dust wall. The heating energy is estimated as $L_{\\rm uv} \\sim 6.3\\times 10^{44}$ erg s$^{-1}$, compatible with either the obscured SED or a low-$\\dot m$ unobscured disk, with implications for UV/optical continuum lags and BLR/DC emission. The campaign also reveals an extended hot-dust component at $\\sim 140-350$ pc, likely related to the host bulge and nuclear star formation, and a spatially resolved narrow-line region traced by [S III] 9531 Å emission. Together, these findings refine our view of the AGN dusty structure, its heating sources, and the connection between nuclear activity and circumnuclear dust on both sub-parsec and kiloparsec scales.

Abstract

The AGN Space Telescope and Optical Reverberation Mapping 2 (STORM 2) campaign targeted Mrk 817 with intensive multi-wavelength monitoring and found its soft X-ray emission to be strongly absorbed. We present results from 157 near-IR spectra with an average cadence of a few days. Whereas the hot dust reverberation signal as tracked by the continuum flux does not have a clear response, we recover a dust reverberation radius of $\sim 90$ light-days from the blackbody dust temperature light-curve. This radius is consistent with previous photometric reverberation mapping results when Mrk 817 was in an unobscured state. The heating/cooling process we observe indicates that the inner limit of the dusty torus is set by a process other than sublimation, rendering it a luminosity-invariant `dusty wall' of a carbonaceous composition. Assuming thermal equilibrium for dust optically thick to the incident radiation, we derive a luminosity of $\sim 6 \times 10^{44}$ erg s$^{-1}$ for the source heating it. This luminosity is similar to that of the obscured spectral energy distribution, assuming a disk with an Eddington accretion rate of $\dot{m} \sim 0.2$. Alternatively, the dust is illuminated by an unobscured lower luminosity disk with $\dot{m} \sim 0.1$, which permits the UV/optical continuum lags in the high-obscuration state to be dominated by diffuse emission from the broad-line region. Finally, we find hot dust extended on scales $> 140-350$ pc, associated with the rotating disk of ionised gas we observe in spatially-resolved [SIII] $λ9531$ images. Its likely origin is in the compact bulge of the barred spiral host galaxy, where it is heated by a nuclear starburst.

Figures (17)

• Figure 1: Near-IR spectra of Mrk 817 obtained with the cross-dispersed instruments TripleSpec on ARC (on 2020 Dec 26; blue solid line), GNIRS on Gemini North (on 2021 Jan 14; red solid line) and SpeX on IRTF (on 2021 Mar 19; black solid line) shown as observed flux versus rest-frame wavelength. Emission lines listed in Table 4 of L08a are marked by dotted lines and labeled; black: permitted transitions, green: permitted Fe II multiplets (not labeled), red: forbidden transitions and cyan: forbidden transitions of iron (those of [Fe II] not labeled).
• Figure 2: The mean spectra for the IRTF (cyan), Gemini North (black) and ARC data sets (red) derived after the application of the PrepSpec photometric correction factors and shown as luminosity versus rest-frame wavelength. The NIFS spectrum is also plotted (blue). The vertical dashed lines indicate the wavelength regions used to derive the spectral light-curves for the $z_s$ band (Fig. \ref{['zbandlcurves']}), the accretion disk (Fig. \ref{['disklc']}) and the dust $H$ and $K$ bands (Fig. \ref{['HKbandlc']}).
• Figure 3: Top two panels: Photometric $g$ and $z_s$ band light-curves during the STORM 2 campaign from 2020 Nov 24 (MJD 59177) to 2022 Feb 24 (MJD 59634) and the beginning of the Extended Campaign. Bottom two panels: Gemini North (black filled circles) and IRTF (black filled triangles) near-IR spectral light-curve around the observed wavelength of $8700$ Å, both original and corrected using PrepSpec photometric correction factors based on the [S III] $\lambda 9531$ narrow line. The original $z_s$ band light-curve (red filled circles) and the $z_s$ band light-curve corrected for a constant host galaxy contribution (green filled circles) to match the HST/STIS spectral fluxes (blue filled squares) are also shown.
• Figure 4: Top two panels: Photometric $g$ and $z_s$ band light-curves during the STORM 2 campaign (2020 Nov 24 to 2022 May 1) and the Extended Campaign. Bottom two panels: Near-IR spectral light-curve in the rest-frame wavelength region of $\lambda=9730-9790$ Å, which is dominated by accretion disk contributions, for the Gemini North (black filled circles), IRTF (cyan filled triangles) and ARC spectra (red filled circles). The NIFS spectrum is also plotted (blue filled square). We attribute the offset to higher fluxes of the ARC spectra relative to the Gemini North spectra and of these relative to the IRTF spectra to differences in the host galaxy light contribution. Subtracting this constant component gives a well-aligned near-IR spectral light-curve (bottom panel). The vertical black dashed lines mark the periods of high- and low-obscuration until the end of the STORM 2 campaign, as defined by Storm2-paper3. The vertical red dotted line marks the date of the X-ray flare (MJD=59329).
• Figure 5: Top two panels: Near-IR spectral light-curves for the rest-frame wavelength regions of $\lambda=1.50-1.55~\mu$m (H band) and $\lambda=2.05-2.10~\mu$m (K band) for the Gemini North (black filled circles), IRTF (cyan filled triangles) and ARC spectra (red filled circles). The WIRC photometry is also plotted (blue filled squares). We attribute the offset of the IRTF spectra to lower fluxes to an additional dust component contributing to the Gemini North and ARC spectra. Subtracting this constant dust component gives well-aligned near-IR spectral light-curves (bottom two panels).