The role of grain size in AGN torus dust models

Abstract: Fits the infrared spectra from the nuclear regions of AGN can place constraints on the dust properties, distribution, and geometry by comparison with models. However, none of the currently available models fully describe the observations of AGN currently available. Among the aspects least explored, here we focus on the role of dust grain size. We offer the community a new spectral energy distribution (SED) library, hereinafter [GoMar23] model, which is based on the two-phase torus model developed before with the inclusion of the grain size as a model parameter, parameterized by the maximum grain size Psize or equivalently the mass-weighted average grain size < P >. We created 691,200 SEDs using the SKIRT code, where the maximum grain size can vary within the range Psize = 0.01 - 10.0um ( < P >= 0.007 - 3.41um). We fit this new and several existing libraries to a sample of 68 nearby and luminous AGNs with Spitzer/IRS spectra dominated by AGN-heated dust. We find that the [GoMar23] model can adequately reproduce up to 85-88% of the spectra. The dust grain size parameter significantly improves the final fit in up to 90% of these spectra. Statistical tests indicate that the grain size is the third most important parameter in the fitting procedure (after the size and half opening angle of the torus). The requirement of a foreground extinction by our model is lower compared to purely clumpy models. We find that 41% of our sample requires that the maximum dust grain size is as large as Psize =10um (< P >= 3.41um). Nonetheless, we also remark that disk+wind and clumpy torus models are still required to reproduce the spectra of a non-negligible fraction of objects, suggesting the need for several dust geometries to explain the infrared continuum of AGN. This work provides tentative evidence for dust grain growth in the proximity of the AGN.