Resolving the terrestrial planet-forming region of HD 172555 with ALMA: I. Post-impact dust distribution

Abstract

Giant impacts between planetary embryos are a natural step in the terrestrial planet formation process and are expected to create disks of warm debris in the terrestrial regions of their stars. Understanding the gas and dust debris produced in giant impacts is vital for comprehending and constraining models of planetary collisions. We reveal the distribution of millimeter grains in the giant impact debris disk of HD 172555 for the first time, using new ALMA 0.87 mm observations at $\sim$80 mas (2.3 au) resolution. We modeled the interferometric visibilities to obtain basic spatial properties of the disk, and compared it to the disk's dust and gas distributions at other wavelengths. We detect the star and dust emission from an inclined disk out to $\sim$9 au and down to 2.3 au (on-sky) from the central star, with no significant asymmetry in the dust distribution. Radiative transfer modeling of the visibilities indicates the disk surface density distribution of millimeter grains most likely peaks around $\sim$5 au, while the width inferred remains model-dependent at the S/N of the data. We highlight an outward radial offset of the small grains traced by scattered light observations compared to the millimeter grains, which could be explained by the combined effect of gas drag and radiation pressure in the presence of large enough gas densities. Furthermore, SED modeling implies a size distribution slope for the millimeter grains consistent with the expectation of collisional evolution and flatter than inferred for the micron-sized grains, implying a break in the grain size distribution and confirming an overabundance of small grains.

Figures (6)

• Figure 1: Left: ALMA 0.87 mm continuum emission from the HD 172555 planetary system (star and disk) obtained by joint imaging of the combined visibility dataset. Right: same as the left panel, but imaged after removing the star from the visibilities, as described in Sect. \ref{['sect:results']}. The symbol in the center notes the position of the star. In both panels, north is up and east is left. Contours are $\pm$[2, 4, 6] $\times$ 10.4 µJy beam$^{-1}$, the RMS noise level. Images are made with a natural weighting and a u-v taper, as described in Sect. \ref{['sect:observations and data prep']}.
• Figure 2: Left column: same as the right panel in Fig. \ref{['fig:all data plus diskonly']}. Middle column: disk-only models created using RADMC-3D and tclean with the best-fit values from the MCMC fitting (see Table \ref{['tab:final values']}), as described in Sect. \ref{['sect:modeling process']}. Right column: residuals from subtracting the model visibilities from the data visibilities, as described in Sect. \ref{['sect:modeling results']}. Top row: Gaussian model and residuals corresponding to that model. Bottom row: power law model and residuals corresponding to that model. All panels are imaged using a natural weighting and a u-v taper, as described in Sect. \ref{['sect:observations and data prep']}, and the contours are $\pm$[2, 4, 6] $\times$ 10.4 µJy beam$^{-1}$, the RMS noise level.
• Figure 3: Left: full multi-wavelength photometry for HD 172555, including new millimeter photometry (colored symbols) from our observations. Right: zoom in from the left panel focusing on ALMA (sub)millimeter data. In both panels, diamonds correspond to the higher spectral setup and squares correspond to the lower spectral setup. Dark orange markers are stellar fluxes, blue markers are disk fluxes, and dark green markers are total (disk and stellar) combined fluxes. Error bars are $1\sigma$. Previous photometric measurements are in black, where circles are detections and downward-facing triangles are upper limits (photometric points are listed in Table \ref{['tab:photometry']}). The best-fit star (orange, solid) and single-component modified blackbody disk models (blue, dash-dotted) are included and obtained using the method from yelverton_statistically_2019.
• Figure 4: Surface density distribution of the millimeter grains from this paper (Gaussian model in blue, power law model in purple) compared to the the surface density distribution from scattered light imaging (orange, engler_detection_2018engler_detection_2018). The orange dashed line is the inner limit of the scattered light observations. All curves have been normalized to 1, and the shaded regions correspond to 1$\sigma$ error.
• Figure 5: One- and two-dimensional posterior probability distributions for all of the MCMC parameters from the Gaussian model (described in Sect. \ref{['sect:physical model']}), where we applied a Gaussian prior to the inclination $i$. Listed above each one-dimensional posterior probability distribution is the $50^{+34}_{-34}$th percentile value of the posterior probability distribution of each parameter marginalized over all other parameters.