Dynamical Mass Constraints on Transition Disk Perturbers with the G23H Catalog

TL;DR

This study leverages calibrated absolute astrometry from Hipparcos and Gaia through the G23H catalog to place dynamical mass constraints on companions in 11 transition-disk systems, addressing biases from disk-induced photocenter shifts. Using orbit modelling with Octofitter and Bayesian model comparison across four models, the authors detect seven companions (three with strong evidence, $BF>20$), including a new constraint for AB Aurigae and MWC 758, and confirm non-detections in several other systems. Cross-model comparisons reveal possible short-timescale astrometric excess noise likely related to inner-disk variability, and the results are contextualized with observed disk cavities and literature companions. They also present Gaia DR4 simulation predictions for PDS 70 and WISPIT 2, illustrating DR4’s potential to probe Jupiter-mass planets at 1–5 au in disk systems and outline the path toward a comprehensive protoplanet demographics census in the coming era. Overall, the work demonstrates the viability and limitations of astrometry-based dynamical mass measurements in protoplanetary disks and provides concrete mass–orbit constraints for several key systems while mapping the future impact of Gaia DR4.

Abstract

We present dynamical mass constraints on perturbers in 11 transition disk systems using a novel combination of calibrated Hipparcos and Gaia absolute astrometry data. Out of the sample of 11, we find support for companions in seven systems, with significant detections in three. These systems are: HD 142527, where we clearly detect the known low-mass stellar companion HD 142527 B; AB Aurigae, where we detect a low-mass stellar or sub-stellar companion; and MWC 758, where we detect a likely sub-stellar companion. We also find strong evidence of companions to HD 97048 and UX Tau A, and moderate evidence for companions to HD 100546 and CQ Tau. In the four systems with non-detections, we find no evidence for companions more massive than $\sim$6 $M_{\mathrm{Jup}}$ with a semi-major axis greater than 3 au for HD 100453, nor for companions more massive than $\sim$2 $M_{\mathrm{Jup}}$ with a semi-major axis greater than 2 au for TW Hya. We also find no evidence for stellar mass companions with semi-major axes between $\sim$3 and $\sim$25 au for both HD 34282 and RY Lup. In addition to our fiducial model, we perform cross validation between astrometry sources. By comparing results across models, we find tentative evidence of a short timescale excess astrometric noise that may impact some protoplanetary disk systems. We conclude with predictions for the prospects of making dynamical mass constraints on protoplanets in protoplanetary disk systems with Gaia data release 4 using detailed simulations of Gaia DR4 data of PDS 70 and WISPIT 2.

Figures (10)

• Figure 1: Companion mass and semi-major axis marginal posterior for the (fiducial) jitter model. Panels are ordered by decreasing Bayes factor (BF). The Bayes factor can can be interpreted as the odds ratio in favour of a single companion model over a zero companion model, assuming both are equally likely a priori. For stars with BF $>$ 3, the gold contour represents the 1$\sigma$ posterior density. For stars with BF $<$ 3, the gold curve represents the 95th percentile upper limit. The shaded region shows 0.75$\times R_{cav,mm}$, an empirical estimate of the gas cavity size 2025AA...698A.102R.
• Figure 2: Posterior samples from the co-planar, jitter model consistent with having cleared the observed cavity (see Section \ref{['sec:cav_clear']} for details). The dots denote the position of the companion on 1 March 2026 for each individual posterior sample, with the gold contours denoting the 1$\sigma$ posterior density. The dots are coloured by mass, along with their corresponding orbits, with a colour scale that is clipped to the 5th and 95th percentile of the mass distribution for visual clarity. The background images show ALMA continuum images of each disk from 2025ApJ...981L..30S, 2021AA...648A..19W, 2019NatAs...3.1109P, 2022ApJ...933L...4C, 2020ApJ...905...89Y and 2018ApJ...860..124D, with their respective beam sizes shown in the bottom left corner and a 10 au scale bar shown in the bottom right corner. The ALMA images are centered arbitrarily. In the top right corner of each panel we show the median plus/minus the 84th/16th percentiles of the mass and semi-major axis distributions for all of the plotted samples. $^*$ We note that we plot all of the posterior samples for HD 142527, because the orbit of HD 142527 B is too compact to explain the size of the observed cavity 2024AA...683A...6N.
• Figure 3: We plot the 95th percentile upper limit of the Gaia jitter parameter, $\sigma_{\mu,jitter}$ as a function of the outer disk inclination, shown by the blue filled circles. We also show the median value and the the 5th percentile, where visible, denoted by the black dot, and the error bars. The lack of clear correlation with inclination indicates that this additional noise is independent of the orientation of the outer disk and thus is unlikely to be related to the outer disk. Alternative possible sources of excess astrometric "noise" could be due to multiple massive companions producing a signal not captured by our single companion models, or possibly due to the signal from a variable, bright inner disk.
• Figure 4: GDR2 and GDR3 constraints and GDR4 predicted constraints for PDS 70. The yellow, and the (semi-transparent) teal and red contours denote the 1$\sigma$ posterior density constraints for models that use the DR2,DR32,DR3,UEVA, DR2,D32,DR3, and the DR2,DR3, subsets of the G23H catalog. The magenta curve shows the 95th percentile upper limit from the UEVA constraint, showing no significant signal. The blue filled curve denotes the 95th percentile upper limit calculated from the GDR4 simulation using the single star noise model. The black filled curve shows the same for the measured UEVA noise model. The dashed grey line denotes the stellar mass of 0.88 $M_{\mathrm{\odot}}$.
• Figure 5: GDR2 and GDR3 constraints and GDR4 predicted constraints for WISPIT 2. The yellow and the magenta curves show the 95th percentile upper limit from the DR2,DR32,DR3,UEVA, and the UEVA constraints, respectively, showing no significant signal. The blue filled curve denotes the 95th percentile upper limit calculated from the GDR4 simulation using the single star noise model. The black filled curve shows the same for the measured UEVA noise model. The dashed grey line denotes the stellar mass of 1.08 $M_{\mathrm{\odot}}$. Note that there is no significant constraint from the GDR2, GDR32 or GDR3 proper motions, so the individual model results are omitted and the DR2,DR32,DR3,UEVA model constraint is due solely to the UEVA.